Professional Documents
Culture Documents
Discriminative Models
The descriptions above are a characterization of generative models of parsing, in which we define a derivational process that produces a tree, then model the
local steps in that derivation with conditional probabilities. In recent years, discriminative parsing methods
have gained popularity (Collins, 2000; Riezler et al.,
2002). These methods need not be probabilistic at all,
and may or may not decompose parses into independent derivation steps. Rather, they decompose structures into features (a distinct sense from the above
usage). Features are structural configurations present
in the parse. The presence of a specific context-free
rewrite can be a feature of a tree, but so can the length
of the yield of the tree, the presence of a specific pair of
words occurring on the trees boundary, or the presence
of a parallel-structured coordination. In discriminative
approaches, we can, in principle, have any features we
like, though, in practice, some kinds of nonlocal features can drastically reduce the efficiency of parameter
estimation and parsing. Estimation procedures for
discriminative methods do not involve counting and
smoothing, but rather involve iteratively running a
parser on the training set, seeing the errors it makes,
and adjusting feature scores so as to improve the model
for the next iteration. Discriminative approaches can
result in better parsing accuracies, since they allow
more sophisticated feature engineering, but they are
often prohibitively slow to train (reranking systems
are a notable exception).
Unsupervised Models
Search Methods
Once a grammar family has been chosen, and a scoring model has been designed, it can be a substantial
challenge to engineer implementations that are simultaneously fast, robust, and accurate. As a result, a
good deal of parsing work has focused on efficiency
issues.
Given unlimited computational resources, one can
find the best-scoring derivation for a sentence by listing
all possible derivations, scoring each one, and recording which one received the highest score. In some cases,
a list-and-score approach is perfectly adequate. In
reranking systems, or with extremely constrained
grammars, there may only be a few dozen candidates.
However, with most grammars, the cost of broad coverage is massive ambiguity; it is not unusual for even
short sentences to have millions of derivations. Therefore, we must find ways of searching for good parses
without actually writing each possibility down.
One general framework for search is known as
state-space search. In state-space methods, we arrange the set of all partial derivations into a huge
tree, in which each node is an incomplete derivation
with one child for each immediate extension of that
derivation. Paths in this tree go from the empty derivation at the root to full parses at the leaves. As we
flesh out those partial derivations, our candidate set
grows rapidly. In beam search, we keep that growing
set small by discarding poor partial derivations early
on, keeping only a few hypotheses alive at any time.
The standard approach is to keep a constant number
of top candidates, or to keep all candidates whose
score is close to the current best candidate. The bestscoring full derivation can of course be mistakenly
discarded early on because it initially looked less
promising than other candidates, giving an effect
analogous to garden pathing in human processing.
Beam search can be sped up by shrinking the beam,
all the way down to greedy search with a beam of size
one, at the cost of degraded accuracy. For many complex models, beam search methods are the best option, since they make no assumption about the kinds
of derivation steps and features in the model.
In many cases, another option, called dynamic programming, is possible. In dynamic programming, we
independently find best parses for various subspans of
the input sentence, then combine those subparses to
form larger parse fragments. Dynamic programs are
typically slower than beam search methods, especially for richer grammar formalisms. Their advantage is
that, while beam search is approximate, dynamic
programming algorithms are guaranteed to return
the parse with the highest score, a property known
as optimality. Dynamic programming algorithms can
also be accelerated, either by approximate methods
(Caraballo and Charniak, 1998) or by an AI technique known as A* search (Klein and Manning,
2003).
Current Research
Current research on statistical parsing spans all of the
aspects above, and includes other topics that are receiving increasing attention. Now that parsers are
finding more applications, it has become clear that
even the best statistical parsers suffer substantially
degraded accuracy when applied to text that is very
different in character from the data (usually newswire) they were trained on. Medical text, web pages,
and spoken language are just a few domains where
this degradation occurs, and techniques for adaptation are just starting to be investigated. Another relatively recent topic of interest is the production of
parsers that can produce semantic structures of some
kind, for example the identification of semantic roles
or the conversion of a parse into predicate-argument
structure (Gildea and Jurafsky, 2002).
Of course, improving the accuracy of parsers is
always of value on the task of parsing the Penn
Treebank, top parsing accuracies are in the very low
90% range. While some parser errors seem difficult to
resolve in an automatic fashion without serious models of semantics and real-world knowledge, most
errors are still the kinds that might reasonably be
improved with better statistical models.
For further reading on statistical parsing methods,
see Manning and Shu tze (1999) and Jurafsky and
Martin (2000).
See also: Categorial Grammars: Deductive Approaches;
Chart Parsing and Well-Formed Substring Tables; Language Processing: Statistical Methods; Parsing and
Grammar Description, Corpus-Based; Part-of-Speech
Tagging; Treebanks and Tagsets; Unification: Computational Issues.
Bibliography
Caraballo S A & Charniak E (1998). New figures of merit
for best-first probabilistic chart parsing. Computational
Linguistics 24, 275298.
career on the area of formal semantics and its association with syntax, pragmatics, and logic, with a special interest in quantification. Her work in the field
has been characterized by active attempts to embrace
and to synthesize diverse works and diverse perspectives such as mathematics and linguistics, formal
semantics and syntax, philosophy and cognitive science, and transformational and non-transformational
References
Akmaijan A & Lehrer A (1976). NP-like quantifiers and
the problem of determining the head of an NP. Linguistic
Analysis 2(4), 395413.
Battye A (1991). Partitive and pseudo-partitive revisited:
reflections on the status of de in French. Journal of
French Languages Studies 1, 2143.
Corbett G (1978). Universals in the syntax of cardinal
numerals. Lingua 46, 355368.
Croft W (2001). Radical construction grammar. Syntactic
theory in typological perspective. Oxford: Oxford University Press.
Delsing L-O (1993). The internal structure of noun phrases
in the Scandinavian languages. Ph.D. diss., University of
Lund.
Greenberg J (1989). The internal and external syntax of
numeral expressions. Belgian Journal of Linguistics 4,
105118.
Heine B & Kuteva T (2002). World lexicon of grammaticalization. Cambridge: Cambridge University Press.
Jackendoff R (1977). X syntax: a study of phrase structure.
Cambridge, MA: The MIT Press.
Part-of-Speech Tagging
T Brants, Google Inc., Mountain View, CA, USA
2006 Elsevier Ltd. All rights reserved.
Introduction
Part-of-speech tagging is the task of determining the
syntactic class (i.e., part of speech) of each word in a
natural language text and assigning it a tag accordingly. The tags generally encode coarse syntactic categories like noun, verb, adjective, etc., as well as
selected sub-categories and morphological features
like number, tense, definiteness, etc. Typical tagsets
contain 20 to 100 categories, and in some cases cover
400 or more categories (see Treebanks and Tagsets).
The following is an example from the Wall Street
Journal portion of the Penn Treebank:
State-of-the-art systems achieve around 97% accuracy on English text for moderately-sized tagsets. Major
paradigms are the following:
. Statistical: Statistical techniques model probability
distributions over tags by using transition probabilities between tags or words and tags, and lexical
probabilities of tags for words. The process tries
k
Q
i1
1 if t T and c matches C
0 otherwise
pcptjcf
c; t
c;t
has an expected value of p (f ) 0.0023 by determining its relative frequency in a corpus. We then require
that our model assign the same value.
The goal is to find the conditional probability distributions p*(t|c) that match the given constraints and
maximize the conditional entropy.
(9) Hp
pcptjclogtjc
c;t
n
1 Y
eli fi c;t
Zl c i1
Classification-Based Tagging
Many different machine learning classifiers have
been explored for part-of-speech tagging. They generally use neighboring words and part-of-speech
tags, and information about possible tags for the
current word as features, make local decisions
for one word at a time, and dont try to estimate
probability distributions for tags.
Memory-Based Learning
n
P
Gfi dxi ; yi
i1
0 if xi yi
1 otherwise
Transformation-Based Tagging
Brills original algorithm of rule application is inherently slow because each rule is compared to each
position in the text, regardless of previous comparisons and previous partial matches, and interactions
between rules might cause multiple changes of the
same tag. Using this observation, Roche and Schabes
(1995) presented an efficient reformulation as a
finite-state transducer.
The resulting tagger operates in time linear to the
length of the text independent of the number of
rules and size of context inspected. In practice, a
speedup by a factor of more than 20 compared to
Brills original tagger is reported.
Classifier Combination
Classifier combinations, also known as classifier
ensembles or stacked classifiers, can lead to significantly improved tagging accuracies. Classifiers are
combined by voting; weighting; or using a secondlevel classifier.
In a heterogeneous combination, several different types of classifiers are combined. Brill and Wu
(1998) combined three taggers (transformation-
NN
CD
NNP
JJ
VB
VBP
(1) tokens
(2) types
(3) low-freq tokens
55.27%
74.72%
79.81%
36.44%
0.43%
0.09%
7.13%
13.54%
13.78%
0.73%
9.25%
5.83%
0.24%
1.20%
0.36%
0.17%
0.69%
0.14%
only 84.1%. Having the tagger assign tags to characters instead of words and using character-based features results in a much higher accuracy of 91.7%.
Combining segmentation and character-based tagging into one maximum-entropy-based model yields
an accuracy of 91.9%. This slightly higher accuracy
comes with the price of 10 times longer processing
times.
Kudo et al. (2004) used Conditional Random
Fields to jointly assign segment boundaries and partof-speech tags to Japanese texts. They reported a
combined segmentation/part-of-speech F-Score of
98.31% on the Kyoto University Corpus (please note
that Chinese and Japanese use different evaluation
metrics). This is significantly higher than results
reported for HMMs, maximum-entropy-based tagging, and rule-based tagging.
Agglutinative languages like Finnish, Hungarian,
Turkish, and Korean create words by combining a
potentially large number of morphemes. For these
languages, the morpheme is a better unit to assign
part-of-speech tags. Assignment on the word-level
would lead to thousands of different tags. These
languages either require an additional step of segmentation, or separate tag assignments for different
morphemes without explicit segmentation.
Hakkani-Tu r et al. (2000) use the notion of inflectional groups as the building blocks of Turkish words.
They use an HMM to identify inflectional groups and
assign part-of-speech, and report an accuracy of
93.95%.
A different approach to tagging agglutinative languages and other languages with potentially huge
tagsets is to systematically reduce the number of
tags used without losing too much information.
Tufis et al. (2000) presented a procedure named
tiered tagging. Several tags of the surface tagset are
combined into one tag of the internal tagset such that
the external tag can be recovered in most cases. They
reduced the tagset from 2148 to 119 tags. Accuracy
with the reduced tagset is 97.49%.
Accuracy
Reference
Model Combination
Model Combination
97.2%
97.2%
97.2%
97.1%
97.1%
96.7%
96.7%
96.6%
96.6%
96.4%
Daelemans et al.,
1996
Conclusions
96.4% and 96.7% on the WSJ corpus. These algorithms have been tested on a large variety of data sets
and languages, generally with good results. Newer,
more complex models achieve accuracies between
96.9% and 97.2%. So far, these models have been
tested on a much smaller variety of data sets. Table 2
lists state-of-the art tagging accuracies on Wall Street
Journal data found in the literature. For comparison,
human accuracy is 98%99%.
Accuracies reported by different authors usually
need to be taken with a grain of salt. The major
source of differences is the use of different corpora
and/or tag sets. Such a difference makes a direct
comparison impossible. Therefore, Table 2 only
includes results on the WSJ portion of the Penn
Treebank. Another major difference is the amount
of training data used. Generally, more data is better, and while learning curves are relatively flat at 1
million words, we still expect improvements when
increasing the data size by one order of magnitude.
Some variation is caused by the way the data is split.
Single splits into one training set and one test set are
especially susceptible to random variation. Performing a 10-fold cross-validation reduces the variation:
the data is split into 10 parts, and the experiment is
repeated 10 times. Each time, 1 part is used for testing, the other 9 for training. Even with this method,
there is room for variation. Splitting into 10 contiguous chunks, then using 9 for training and 1 for testing
usually yields slightly worse results than splitting sentence-wise, i.e., take 9 sentences for training and 1 for
testing. Cross-validation also allows to accurately
assess the variation in results.
An occasional practice is to test many parameter
settings and then report the best results. This yields
biased results since it uses parameters that are
optimized on the test set. A better experimental
Acknowledgments
I would like to thank Sabine Brants, Peter Dienes,
Alex Franz, Julian Kupiec, Dekang Lin, and Wojciech
Skut for valuable discussions on the topics covered,
for deciding on and structuring the content, and for
help in preparing this article.
Bibliography
Abney S P, Schapire R E & Singer Y (1999). Boosting
applied to tagging and PP attachment. In Proceedings
of the Joint SIGDAT Conference on Empirical Methods
in Natural Language Processing and Very Large Corpora
(EMNLP/VLC-99), College Park, MD, USA. 3845.
Brants T (2000). TnT a statistical part-of-speech tagger.
In Proceedings of the Sixth Conference on Applied Natural Language Processing ANLP-2000, Seattle, WA, USA.
224231.
Brill E & Wu J (1998). Classifier combination for
improved lexical disambiguation. In Proceedings of
the 17th International Conference on Computational
Linguistics (COLING-ACL-98), Montreal, Canada.
191195.
Brill E (1995). Transformation-based error-driven learning
and natural language processing: a case study in part-ofspeech tagging. Computational Linguistics 21(4),
543566.
Chanod J-P & Tapanainen P (1995). Tagging French: comparing a statistical and a constraint-based method. In
Proceedings of Seventh Conference of the European
Chapter of the Association for Computational Linguistics EACL-95, Dublin, Ireland. 149156.
Church K W (1988). A stochastic parts program and
noun phrase parser for unrestricted text. In Proceedings
of the Second Conference on Applied Natural
Language Processing ANLP-88, Austin, Texas, USA.
136143.
Collins M (2002). Discriminative training methods for
hidden Markov models: theory and experiments with
perceptron algorithms. In Proceedings of the Conference
on Empirical Methods in Natural Language Processing
(EMNLP-02), Philadelphia, PA. 18.
Cutting D, Kupiec J, Pedersen J & Sibun P (1992).
A practical part-of-speech tagger. In Proceedings of the
third Conference on Applied Natural Language
Processing (ACL), Trento, Italy. 133140.
Daelemans W, Zavrel J, Berck P & Gillis S (1996). MBT:
A memory-based part of speech tagger-generator. In
Proceedings of the Fourth Workshop on Very Large
Corpora, Copenhagen, Denmark. 1427.
DeRose S J (1988). Grammatical category disambiguation
by statistical optimization. Computational Linguistics
14(1), 3139.
Elworthy D (1994). Does Baum-Welch re-estimation help
taggers? In Proceedings of the Fourth Conference
on Applied Natural Language Processing ANLP-94,
Stuttgart, Germany. 5358.
Erjavec T & Dzeroski S (2004). Machine learning of morphosyntactic structure: lemmatizing unknown Slovene
words. Applied Artificial Intelligence 18, 1741.
Hajic J, Krbec P, Kveton P, Oliva K & Petkevic V (2001).
Serial combination of rules and statistics: a case study
in Czech tagging. In Proceedings of the 39th Annual
Meeting of the Association of Computational Linguistics
(ACL-01), Toulouse, France. 260267.
Hakkani-Tu r D Z, Oflazer K & Tu r G (2000). Statistical morphological disambiguation for agglutinative
Bibliography
Eric Partridge; Unique approach to origins of words (obituary). The Times (London), Nov. 16, 1979, 4.
Green J (2004). Partridge, Eric Honeywood. Oxford
dictionary of national biography. Oxford: Oxford University Press.
Shenker I (1979). Eric Partridge, expert on English and
lover of its quirks, is dead. New York Times June 2,
1979, 1.
Current Research
Current research on statistical parsing spans all of the
aspects above, and includes other topics that are receiving increasing attention. Now that parsers are
finding more applications, it has become clear that
even the best statistical parsers suffer substantially
degraded accuracy when applied to text that is very
different in character from the data (usually newswire) they were trained on. Medical text, web pages,
and spoken language are just a few domains where
this degradation occurs, and techniques for adaptation are just starting to be investigated. Another relatively recent topic of interest is the production of
parsers that can produce semantic structures of some
kind, for example the identification of semantic roles
or the conversion of a parse into predicate-argument
structure (Gildea and Jurafsky, 2002).
Of course, improving the accuracy of parsers is
always of value on the task of parsing the Penn
Treebank, top parsing accuracies are in the very low
90% range. While some parser errors seem difficult to
resolve in an automatic fashion without serious models of semantics and real-world knowledge, most
errors are still the kinds that might reasonably be
improved with better statistical models.
For further reading on statistical parsing methods,
see Manning and Shutze (1999) and Jurafsky and
Martin (2000).
See also: Categorial Grammars: Deductive Approaches;
Chart Parsing and Well-Formed Substring Tables; Language Processing: Statistical Methods; Parsing and
Grammar Description, Corpus-Based; Part-of-Speech
Tagging; Treebanks and Tagsets; Unification: Computational Issues.
Bibliography
Caraballo S A & Charniak E (1998). New figures of merit
for best-first probabilistic chart parsing. Computational
Linguistics 24, 275298.
career on the area of formal semantics and its association with syntax, pragmatics, and logic, with a special interest in quantification. Her work in the field
has been characterized by active attempts to embrace
and to synthesize diverse works and diverse perspectives such as mathematics and linguistics, formal
semantics and syntax, philosophy and cognitive science, and transformational and non-transformational
Bibliography
Borschev V & Partee B H (2002). The Russian genitive of
negation in existential sentences: The role of Theme-Rheme
structure reconsidered. In Hajicova E, Sgall P, Hana J &
Hoskovec T (eds.) Travaux du Cercle Linguistique de Prague (novelle serie), v. 4. Amsterdam/Philadelphia: John
Benjamins Publishing Company. 185250.
Partee B H (1971). On the requirement that transformations
preserve meaning. In Fillmore C J & Langendoen D T
(eds.) Studies in linguistic semantics. Austin, TX: Holt,
Rinehart, and Winston. 121.
Partee B H (1978). Fundamentals of mathematics for linguists. Dordrecht: Reidel.
Partee B H (1979). Subject and object in Modern English.
New York: Garland Publishing, Inc. New York. [Publication of Hall, B. H., 1965, MIT dissertation.].
Partee B H (1996). The development of formal semantics
in linguistic theory. In Lappin S (ed.) The handbook
of contemporary semantic theory. Oxford: Blackwell.
1138.
Partee B H (2004). Compositionality in formal semantics:
selected papers by Barbara H. Partee. Oxford: Blackwell
Publishing. [Introduction is autobiographical.]
Stockwell R P, Schachter P & Partee B H (1972). The
major syntactic structures of English. Austin, TX: Holt,
Rinehart, and Winston.
research that explored second dialect acquisition, social networks and their linguistic consequences, and
the impact of linguistic accommodation. Trudgills
(2000) broad, English-based studies provided the
foundation for comprehensive surveys of the entire
field of sociolinguistics. As is the case in the United
States, efforts among linguists throughout the world
to serve as social advocates include direct and indirect
approaches.
Another dimension of linguistic advocacy is related
to sex. Eckert and McConnell-Ginet (2003), Holmes
and Meyerhoff (2003), and others analyzed actual
and idealized differences in mens and womens language, and the usage of language among women
when men are absent. The advocacy dimension that
these works share sought to advance greater opportunities and equality for women worldwide. They
cut across race, region, class, and education, which
are often discussed in isolation of matters pertaining
exclusively to women.
Skutnabb-Kangas and Phillipson (1994), Alexander (1989), Gibson (2004) and Mesthrie (1992)
sought to address human linguistic rights as a generic subject, or as related to the specific sociolinguistic
and political circumstances of specific nations. Complementary studies of linguistic foundations of social
strife were developed by Joshua Fishman, growing
directly from his pioneering studies of the sociology
of language.
Additional linguistic research devoted to public
advocacy of various kinds, say, attending to education, women, or uses of language in the workplace,
will often have an explicit regional orientation. This
survey does not adequately reflect efforts by scholars who study their local linguistic community (or
communities) with sensitivity to their immediate
sociohistorical circumstances.
Pennycooks (1998) studies of language usage in
Australia and elsewhere had relevance to other
nations that are post-colonial with populations from
diverse backgrounds. Similarly, Bortonis (1985) linguistic studies in Brazil drew attention to the special
plight of indigenous people in the wake of European
colonization and its corresponding educational consequences. Many other studies of indigenous people
strove to advance linguistic dignity with greater educational and social opportunities for people who
were, or remain, marginalized if not displaced in
their native homeland. These concerns are strongly
evident in societies that are former European colonies, where European languages continue to serve
dominant institutional functions while indigenous
languages are often subordinate.
Differences in social opportunity become pronounced when viewed in terms of the distribution of
various social services that are provided either publicly or privately in nations throughout the world.
Linguistic research has also observed differential access to medical treatment or justice based on proficiency, or the lack thereof, of the dominant standard
language in a given society.
Medical applications of linguistics also fall within the realm of advocacy research, and Labov and
Fanshels (1977) Therapeutic discourse, which was
complemented by Ferrarras (1994) Therapeutic
ways with words, was illustrative of analyses that
employed discourse analyses to the benefit of medical
application in general, and therapeutic application
quite specifically. Studies of doctor/patient interaction
are pervasive, as are studies of language usage in the
courts. Policy implications abound from these efforts,
because they frequently expose the relative linguistic
strengths and weaknesses that exist within a society,
particularly if that society seeks to serve populations
that do not share a common language or culture.
Prestons (1989) studies of dialect perceptions
emphasized another reality; namely, that people
tend to judge others linguistic behavior based on
egocentric impressions that are reflective of personal
linguistic exposure. Long before Preston discovered
this among speakers of English throughout the United
States, Lambert (1972) and Tucker and Lambert
(1972) demonstrated that bilingual and bidialectal
attitudes toward speakers of other languages or dialects are firmly entrenched, and they have real consequences for people who are perceived to be more or
less qualified for jobs or educational opportunities
based on their speech.
Research by Baugh (2003) and Smalls (2004) explored the consequences of linguistic profiling and
other forms of linguistic discrimination in U.S. housing markets, where speakers of minority dialects or
languages have been denied access to fair housing
or fair lending over the telephone. Whereas the courts
in the United States are well equipped to prosecute
cases where alleged racial discrimination is based
on face-to-face encounters, Smalls (2004) demonstrated that U.S. courts are less capable of prosecuting cases where discrimination takes place during
telephone conversations. Complementary dialectology studies illustrate that stereotypical names associated with particular racial groups within the United
States result in different opportunities in the workplace. For example, identical resumes were submitted
to prospective employers with only the name of
the applicant changed. Jamal is less likely to get a
job than is Greg, and Lakisha is less likely to get
a job than is Emily; that is, with all other professional
and personal attributes being strictly controlled by
experimental design.
Bibliography
Abrahams R (1985). Afro-American folktales: stories from
Black traditions in the New World. New York: Pantheon
Books.
Alexander N (1989). Language policy and national unity in
South Africa/Azania: an essay. Cape Town: Buchu Books.
Baugh J (2000). Racial identification by speech. American
Speech 75(4), 362364.
Baugh J (2003). Linguistic profiling. In Makoni S,
Smitherman G, Ball A F & Spears A K (eds.) Black
linguistics: language, society, and politics in Africa
and the Americas. London/New York: Routledge Press.
155168.
Bernstein B (1971). Class, codes and control. London:
Routledge.
Bernstein B (1975). Class and pedagogies, visible and invisible. Paris: Organization for Economic Co-operation and
Development.
Bortoni S R (1985). The urbanization of rural dialect
speakers: a sociolinguistic study in Brazil. Cambridge;
New York: Cambridge University Press.
Chambers J & Trudgill P (eds.) (1998). Dialectology (2nd
edn.). Cambridge; New York: Cambridge University
Press.
Cheshire J (1982). Variation in an English dialect: a sociolinguistic study. Cambridge; New York: Cambridge
University Press.
Eckert P & McConnell-Ginet S (2003). Language and gender. Cambridge; New York: Cambridge University Press.
Relevant Websites
http://www.ncsu.edu.
http://www.ncte.org.
Relevant Websites
http://www.ncsu.edu.
http://www.ncte.org.
la difficulte fondamentale en fait en traduction automatique . . . . The particle alors signals that the speaker is about to begin, while the particle en fait [in fact]
highlights la difficulte fondamentale. The spoken text
was some 30% longer than the written text.
Both writers and speakers use discourse particles,
but different particles occur in unplanned speech/
conversation and in various types of written text.
In particular, conversation lends itself to the use of
particles signaling a given speakers attitude to
information conveyed by the previous speaker. The
interactive nature of conversation leads to the use of
particles signaling the receipt of information or
checking whether information has been received.
Such particles occur regularly in conversation but
are particularly frequent in the task-related dialogues
studied by Anderson et al. (1991). The task of the
participants in these dialogues was to cooperate in
drawing routes on maps. The instruction givers had
maps with routes and sets of landmarks; the instruction followers had maps with the start of routes and
sets of landmarks; the two sets of landmarks were not
identical. Checking and back-channel particles such
as right? and OK? with different pitch patterns were
particularly frequent when the participants were
hidden from each other by a screen.
Checking and back-channel particles aside, particles occur regularly in different types of text. According to Biber et al. (1999: 880), linking adverbs are
most frequent in academic writing (7100 per one
million words), slightly less frequent in conversation (5900 per one million words), and relatively
infrequent in fiction and newspapers (2200 and
1800 per one million words, respectively). The higher
figure for conversation results from the intensive use
of so and then, which are infrequent in academic
writing. In contrast, academic writing offers a wide
range of linking particles, such as for example, in
addition, nevertheless, furthermore, and hence. These
were infrequent in the corpus of conversation analyzed by Biber et al. (1999). According to them,
given what is known about the effect of prolonged
formal education on peoples speech, it can safely be
supposed that such linking particles are very rare
in most spontaneous conversation; for example,
they were not found in the conversations analyzed
by Miller and Weinert (1998), even in conversation
between mature university students.
Can generalizations be made about where discourse particles occur in clauses/utterances? Schiffrin
(2000: 57) defined discourse particles or markers as
sequentially dependent elements that bracket units of
talk; this definition covers items such as not only . . .
but also. Schiffrin elaborated the definition as follows: i.e., [discourse markers are] non-obligatory
Conclusion
Bibliography
Anderson A H, Bader M, Bard E G et al. (1991). The
HCRC map task corpus. Language and Speech 34,
351366.
Biber D, Johansson S, Leech G et al. (1999). Longman
grammar of spoken and written English. London:
Longman.
Brinton L (1996). Pragmatic markers in English grammaticalization and discourse functions. The Hague: Mouton
de Gruyter.
Fernandez M M J (1994). Les particules enonciatives. Paris:
Presses Universitaires de France.
Fraser B (1998). Contrastive discourse markers in English.
In Jucker A H & Ziv Y (eds.) Discourse markers: descriptions and theory. Amsterdam: John Benjamin. 302306.
Halliday M A K & Hasan R (1976). Cohesion in English.
London: Longman.
Meyerhoff M (1994). Sounds pretty ethnic, eh?: a pragmatic particle in New Zealand English. Language in
Society 23, 367388.
Miller J & Weinert R (1998). Spontaneous spoken language: syntax and discourse. Oxford: Clarendon Press.
Ostman J (1981). You know: a discourse functional
approach. Amsterdam: John Benjamin.
Palmer F R (2001). Mood and modality. Cambridge:
Cambridge University Press.
Quirk R & Greenbaum S (1973). University grammar of
English. London: Longman.
Schriffrin D (1987). Discourse markers. Cambridge:
Cambridge University Press.
Schiffrin D (ed.) (2000). The handbook of discourse analysis. Oxford: Blackwell.
Unger C (1996). The scope of discourse connectives: implications for discourse organisation. Journal of Linguistics
32, 403448.
218 Partitives
Partitives
M Koptjevskaja-Tamm, Stockholm University,
Stockholm, Sweden
2006 Elsevier Ltd. All rights reserved.
In traditional linguistics, primarily in the IndoEuropeanistic tradition, the term partitive is normally associated with partitive (meanings/uses of)
genitives, which include (a) reference to body parts
and organic parts of objects, e.g., the roof of the
house, the lions head; (b) reference to a set from
which a subset is selected by means of various nonverbal words, e.g., the best among the Trojans, three
of the boys, a section of the barbarians; (c) quantification, e.g., an amphora of wine, dozens of soldiers;
and (d) reference to partial objects of certain verbs
(such as to eat, to drink, etc.), normally alternating
with accusatives, primarily in Classical Greek, Gothic
and Old High German, Sanskrit, and Balto-Slavic.
The idea of partiality, to which the term partitive
owes its name, grows more and more vague as we
proceed along this list. The last two constructions do
not in fact refer to a part in any reasonable sense, since
there is no well-defined whole to which it could relate.
The partitive case is, together with the nominative, accusative, and genitive, one of the four main
grammatical cases in the Finnic languages. It continues the proto-Uralic separative (ablative case) but
has more or less lost its original uses in Finnic. Its
central functions overlap with those of the partitive
genitives and include marking partial objects and
subjects, complements to nominal quantifiers, and
to numerals under certain syntactic conditions. The
functions of the partitive article in French (and in
some other Romance varieties) also include marking
partial objects and subjects.
will be called quantified. Although the same quantifiers may appear in both types of NPs, their role is
different. Thus, PCs involve a presupposed set of
items or a presupposed entity referred to by one of
the nominals (the cake, Marys books), and the quantifier indicates a subset or a subpart which is selected
from it. In a pseudo-partitive nominal construction
(PPC), the same word merely quantifies over the kind
of entity (tea, books) indicated by the other nominal.
Swedish, along with many other languages, makes
a clear distinction between the two constructions:
the quantified in PCs is marked with the preposition
av, whereas PPCs merely consist of two juxtaposed
nominals.
(1a) en
kopp
av de-t
a:COM
cup
of the-NEUT.SING
god-a
te-t
good-DEF tea-DEF.N.SING
a cup of the good tea
(1b) en
kopp te
a:COM cup
tea
a cup of tea
Partitives 219
Ahmet
[sarab-in yari-sin-i]
Ahmet
wine-GEN half-3SING.POSS-ACC
ic-ti
drink-PAST
Ahmet drank half of the wine
For PPCs, the cross-linguistically dominating technique consists in merely juxtaposing the nominal
quantifier and the quantified, as in the Swedish example (1b) above. Since PCs normally involve overt
markers, in these cases there will be a clear contrast
between PCs and PPCs. However, in many languages,
PPCs also involve a construction marker for relating
the nominal quantifier and its complement to each
other either the same as or different from the one
used in PCs.
Although languages tend to have one standard
PC and one standard PPC, they may occasionally
show other, more marginal patterns. These are often
restricted to certain semantic subgroups of nominal quantifiers and/or special contexts of use, as
the Turkish possessive-like PC in example (3). In
Swedish, a group of students can be expressed by
the juxtapositional construction en grupp studenter
or by two constructions involving prepositions: en
220 Partitives
Headedness in Pseudo-Partitive
Constructions
The structure of partitive, and especially pseudo-partitive, constructions, has been debated by various
syntactic theories, in particular, the questions of headedness and constituency (Akmaijan and Lehrer, 1976;
Jackendoff, 1977; Selkirk, 1977; Lo bel, 1986, 2000;
Battye, 1991; Delsing, 1993; Vos, 1999; Kinn, 2001).
The facts are often controversial.
Pseudo-partitive nominal constructions with overt
markers superficially look like typical asymmetrical
head-dependent structures, primarily like (possessive)
noun phrases. Thus, the quantified is marked with an
inflectional or adpositional marker typical of marking dependents, (see examples (4a) and (4b)). In languages with case, it is the quantifier that receives the
morphological case marking appropriate to the slot
filled by the whole construction, i.e., is the morphosyntactic locus of the construction (Zwicky, 1985;
see example (4c)). These marking facts point out the
quantifier as the head and the quantified as the dependent in such constructions (examples 4a to 4c are
from Russian).
(4a) bokal vin-a
glass
wine-GEN
a glass of wine
(4b) bokal Petr-a
glass
Peter-GEN
Peters glass
(4c) smes at
s
bokal-om
mix.INF
with
glass-INSTR
to mix with a glass of wine
vin-a
wine-GEN
Juxtapositional PPCs work differently. In some languages, illustrated by modern Greek in (5), both the
quantifier and the quantified agree in case.
(5a)
[e na
kilo
kafe s]
[one.NEUT.NOM kilo.NOM coffee.NOM]
kostzi
epta
dola ria
cost.PRES.3SING seven
dollars.ACC
One kilo of coffee costs seven dollars
(5b) i
aksa
[eno s
kilu
art.FEM.NOM value.NOM
[one.N.GEN kilo.GEN
kafe ]
ne
epta
dola ria
coffee.GEN] be.PRES.3SING seven
dollars.ACC
The price of one kilo of coffee is seven dollars
References
Akmaijan A & Lehrer A (1976). NP-like quantifiers and
the problem of determining the head of an NP. Linguistic
Analysis 2(4), 395413.
Battye A (1991). Partitive and pseudo-partitive revisited:
reflections on the status of de in French. Journal of
French Languages Studies 1, 2143.
Corbett G (1978). Universals in the syntax of cardinal
numerals. Lingua 46, 355368.
Croft W (2001). Radical construction grammar. Syntactic
theory in typological perspective. Oxford: Oxford University Press.
Delsing L-O (1993). The internal structure of noun phrases
in the Scandinavian languages. Ph.D. diss., University of
Lund.
Greenberg J (1989). The internal and external syntax of
numeral expressions. Belgian Journal of Linguistics 4,
105118.
Heine B & Kuteva T (2002). World lexicon of grammaticalization. Cambridge: Cambridge University Press.
Jackendoff R (1977). X syntax: a study of phrase structure.
Cambridge, MA: The MIT Press.
Part-of-Speech Tagging
T Brants, Google Inc., Mountain View, CA, USA
2006 Elsevier Ltd. All rights reserved.
Introduction
Part-of-speech tagging is the task of determining the
syntactic class (i.e., part of speech) of each word in a
natural language text and assigning it a tag accordingly. The tags generally encode coarse syntactic categories like noun, verb, adjective, etc., as well as
selected sub-categories and morphological features
like number, tense, definiteness, etc. Typical tagsets
contain 20 to 100 categories, and in some cases cover
400 or more categories (see Treebanks and Tagsets).
The following is an example from the Wall Street
Journal portion of the Penn Treebank:
Bibliography
Eric Partridge; Unique approach to origins of words (obituary). The Times (London), Nov. 16, 1979, 4.
Green J (2004). Partridge, Eric Honeywood. Oxford
dictionary of national biography. Oxford: Oxford University Press.
Shenker I (1979). Eric Partridge, expert on English and
lover of its quirks, is dead. New York Times June 2,
1979, 1.
Pashto 231
Pashto
There is also a large diaspora in the Gulf countries,
particularly in Dubai, and in Europe, the United
States, and Australia.
Dialectology
Table 1 Dialects
Zone C
x
g
Standard
Pashto
Afgh
Pak
Nonstandard
Pashto
Afgh
Pak
Z
Kandahar
Quetta
Zone A
x [c]
g [_]
x
g
Djalalabad
Peshawar
Paktya
Waziristan
Bannu
Zone C
maWrebi
S
Z
Ghazni
Zone B
Zone B
mandaney
Zone A
maSreqi
father/pla:r/
mother/mor/
daughter /lur/
father /plor/
mother /mer/
daughter /lir/
232 Pashto
Basic Phonology
Pashto is a language with free accentuation.
A remarkable feature of Pashto is a series of
retroflex consonants (/</, /B/, //, /Z/, /0/), which is
exceptional in Iranian languages; Pashto also has
many word-initial clusters that cannot exist in Persian
Table 3.
Table 2 Consonants
Bilabial
Dental/Alveolar
Retroflex
Velar
Plosive
Affricate
<
qb
Fricative
t
ts
0
s
xa
S
0
Nasal
Liquid
Uvular
dz
d
z
ga
8
r
Semivowels
a
Cf. Dialectology.
In italics: the elegant phonemes.
These phonemes are not native pashto sounds. They occur in the speech of educated speakers only (in Arabic and Persian loan words).
/q/ varies with /k/ in a stylistically determinated alternation, /f/ with /p/ and /h/ lengthening a preceding vowel /a/ with zero.
b
Pashto 233
Basic Morphology
Nouns
Table 3 Vowels
Front
Central
Back
u
e
o
e
a
a:
NP
1 sing
2 sing
3MASC
3 sing
3FEMsing
1 pl
2 pl
3.MASC.PL
3.PL
3 FEM.PL
Directional
Personal ending
NOM
OBL
(OBL)
Present
ze
te
day
-da:
ma:
ta:
de
de
-me
-de
ra:
dar
-ye
war
-i
mung
ta:se / ta:so
-mo
-mo
ra:
dar
u
ey
duy
-ye
war
Past
em(a)
e
e/ay/
a
e
Strong pronouns
Weak pronouns
234 Pashto
Table 5 Nominal determination
Head modifier
Head noun
zye8
de ma:zi"gar
de ma:zi"gar
of aternoon.OBL
zye8
yellow.NOM
"weli"del-a
Vo
PERFsee.PAST-3FEM.sing
FEM.NOM
"weli"del-a
Vo
PERF-see.PAST-3
FEM.sing
ENCL.1.sing
I saw this woman
(3) "weme-li"del-a
PERF-s-V-o
PERF-ENCL.1.sing-see.PAST-3FEM.sing
I saw her
Ergativity
lmar
lmar
lmar
sun.NOM
The construction is accusative in the present because the subject of the transitive verb behaves in
the same way as the subject of the intransitive verb.
In example (5), it is in the direct (nominative) case
and the verb agrees with it.
The construction is ergative in the past because the
object of the transitive verb behaves in the same way
as the subject of the intransitive verb. Thus, in examples (1)(3), it is the object term that is in the direct
case and the subject term that is in the oblique case.
Moreover, the verb agrees with this object, whether it
is given, as in (1) and (2), or not (3).
(4) (ze)
j-em [tl-em]
S
Vs
me.NOM go.PRES-1/sing [go.PAST]
It is I who am going [was going]
(5) (ze)
S
me.NOM
win-em
Vs
voir.PRES-1/sing
Antiimpersonal Verbs
Pashto 235
Table 6 Landey
l
(7) (ze)
ta:
S
O
me.NOM you.OBL
It is I who see you
win-em
Vs
see.PRES-l/sing
The Landey
It is impossible to talk about the Pashtun world without mentioning a popular poetical genre: the landey,
literally short. Often sung, their rhythm is invariable; every one knows a number of landeys and is able
to compose new ones (accented syllables are in bold
(Table 6)).
48
4 8 12
See also: Afghanistan: Language Situation; Avestan;
Balochi; Ergativity; IndoIranian; Iranian Languages;
Lahnda; Pakistan: Language Situation; Perfectives, Imperfectives, and Progressives; Persian, Modern; Urdu.
Bibliography
Bellew H W (1867). A grammar of the Pukkhto or Pukshto
language. London: (Reprint 1983) Peshawar.
Biddulph C E (1890). Afghan poetry of the seventeenth
century, being selections from the poems of Khushal
Khan Khatak. London.
Caroe O (1965). The Pathans. London: Macmillan Press.
Darmesteter J (18881890). Chants populaires des Afghans.
Paris: (Reprint, Socie te Asiatique, collection des ouvrages
orientaux, 2 se rie, Amsterdam: Philo Press).
Elfenbein J (1984). The Wanetsi Connexion. JRAS 5476,
229241.
Grierson G A (1921). Linguistic Survey of India, vol. 10.
Calcutta.
Grammatical voice typologies often establish a primary opposition between active and passive voice, with
the middle voice as an intermediate category. Syntactically, a passive construction is a detransitivized
counterpart of an active construction. But semantically, the passive retains the same number of thematic
roles as the active, typically with a change in relative
prominence. The impersonal constructions found in
Balto-Finnic, Celtic, and in various Balto-Slavic languages form part of a personal-impersonal opposition
that is orthogonal to the active-passive contrast. Although impersonals may serve the same communicative function as the passive, they represent a
distinctive grammatical strategy for suppressing reference to the subject. Like the periphrastic counterparts formed with English one, French on, or
German man, morphological impersonals are (i) not
syntactically detransitivized, and retain direct objects,
(ii) may often be formed from unaccusative verbs,
which lack passive counterparts, and (iii) tend to
imply an indefinite human subject, which cannot
usually be specified further by an agentive oblique.
Passives
Following Jespersen (1924), it is customary to describe the surface subject of the active construction
as the logical subject and the surface object of the
active as the logical object. In an intransitive personal passive corresponding to an active transitive verb,
the logical object is realized as the surface subject of
the passive. The passive verb cannot govern a direct
object (unless the corresponding active is ditransitive), though the participant role associated with the
logical subject remains implicit, and may usually be
expressed as an oblique dependent. The German
examples in (1) illustrate these properties of the
personal passive. The logical subject that is realized
as the surface subject in the active (1a) corresponds to
the optional oblique in the passive (1b). The logical
object is realized by the accusative object in (1a) and
by the nominative subject in (1b).
(1a) Die Kinder haben den Igel gefuttert.
the.NOM children have the.ACC hedgehog fed
The children have fed the hedgehog.
(1b) Der Igel wurde (von den Kindern) gefuttert.
the.NOM hedgehog was by the.DAT children.DAT
fed
The hedgehog was fed by the children.
passives tend to perform slightly different communicative functions. Both constructions may background or downgrade the logical subject relative
to the activity denoted by the verb. The prominence
of the activity is particularly enhanced if no other
dependent is expressed, as in the case of subjectless
passives without agentive obliques. The personal passive can also foreground the logical object. In English, for example, a personal passive with an
agentive oblique may be used to alter theme-rheme
relations that would be regulated in other languages
by word order alternations within active constructions. The subjectless passive lacks this function, but
can be used instead to express indefinite or generic
statements.
Debates about the universal syntactic properties
of passives have turned largely on which properties
of personal passives are regarded as definitional,
and which are either merely characteristic or else
consequences of the definitional properties. RG
accounts such as Perlmutter and Postal (1984) identify object promotion as the primary effect, which was
taken to entail the demotion of the logical subject.
The traditional alternative articulated in Comrie
(1977), treated subject demotion as the primary effect
of passivization, with promotion as an opportunistic
side-effect. Since personal passives exhibit both demotion and promotion, they provide no basis for
deciding between these choices. Subjectless passives
have played a pivotal role in this debate, because
they exhibit the same form variation and subject demotion as personal passives, but show no evidence of
promotion. As Comrie (1977) argued, any passive
rule that is meant to apply to personal passives like
(1) and to subjectless passives like (2) would have to
demote subjects, not promote objects, since there is
no obvious object to promote in (2a). The RG response involved the promotion of non-obvious
(dummy) objects to induce demotion in cases
like (2b).
Unaccusativity
Impersonals
In some languages, one finds a distinctive type of
impersonal construction occupying the communicative niche associated with passive constructions. This
construction is obligatorily subjectless and usually
receives an active indefinite interpretation, in which
the subject is construed as referring to an unspecified
human subject, or to people in general. The functional overlap with passives often encourages a passive
classification in theoretical descriptions, and somewhat less often in the specialist and pedagogical
literature. However impersonal constructions differ
from passives in a number of significant respects,
which are summarized in Blevins (2003). Perhaps
the most important is the fact that they may be
based on unaccusative verbs, and do not permit agentive obliques. Yet unlike Balto-Finnic or Celtic, each
verb has a single no/to form, which receives a past
interpretation.
(4a) Gazete czytano.
newspaper.FEM.ACC read.PAST.IMP
One/they read the newspaper.
(4b) Tu sie pije wo dke .
here REFL drink.3SG vodka.ACC
One drinks vodka here. (Rothstein, 1993: 712)
cusativity.
Bibliography
Babby L H (1989). Subjectlessness, external subcategorization and the Projection Principle. Zbornik za Filologiju i
Linguvistiku 32, 740. Reprinted in Journal of Slavic
Linguistics 10, 341388.
Billings L & Maling J (1995). Accusative-assigning
participial no/to constructions in Ukrainian, Polish and
neighboring languages: an annotated bibliography.
Parts 1 & 2. Journal of Slavic Linguistics 3(1), 177217
and 3(2), 396430.
Passy is best remembered as the founder of the International Phonetic Association. His interests included not
only applied phonetics (language pedagogy and reading
instruction), but also descriptive linguistics, historical
sound change, and basic phonological theory.
Born in Versailles on January 13, 1859, Passy was
educated at home by visiting teachers, governesses,
and his wealthy parents. His father, Fre de ric Passy,
was an economist, advocate for international arbitration, and first (co-)recipient of the Nobel Peace Prize.
Passy and his siblings learned not only French but also
English, German, and Italian. He attended Colle`ge
Sainte Barbe, and although he easily passed his baccalaure at exams at ages 16 and 17, he failed his
licence three times before passing, since the subjects
bored him. As an option to avoid military service
because he was a pacifist, Passy contracted to teach
English and some German for 10 years in various
public schools, colleges, and principally, the Teacher
Training College at Auteuil. From 1879 to 1885, he
studied phonetics on his own, reading Sweet, Sievers,
Vie tor, and Jespersen. At the E cole des Hautes E tudes,
he studied Sanskrit and Gothic and Old High German
under Ferdinand de Saussure (18851887).
Passy gained his doctorat e`s lettres in 1891 with his
secondary thesis on the phonetics of modern Icelandic
Passy is best remembered as the founder of the International Phonetic Association. His interests included not
only applied phonetics (language pedagogy and reading
instruction), but also descriptive linguistics, historical
sound change, and basic phonological theory.
Born in Versailles on January 13, 1859, Passy was
educated at home by visiting teachers, governesses,
and his wealthy parents. His father, Frederic Passy,
was an economist, advocate for international arbitration, and first (co-)recipient of the Nobel Peace Prize.
Passy and his siblings learned not only French but also
English, German, and Italian. He attended Colle`ge
Sainte Barbe, and although he easily passed his baccalaureat exams at ages 16 and 17, he failed his
licence three times before passing, since the subjects
bored him. As an option to avoid military service
because he was a pacifist, Passy contracted to teach
English and some German for 10 years in various
public schools, colleges, and principally, the Teacher
Training College at Auteuil. From 1879 to 1885, he
studied phonetics on his own, reading Sweet, Sievers,
Vietor, and Jespersen. At the Ecole des Hautes Etudes,
he studied Sanskrit and Gothic and Old High German
under Ferdinand de Saussure (18851887).
Passy gained his doctorat e`s lettres in 1891 with his
secondary thesis on the phonetics of modern Icelandic
He was also an active participant in the Reform Movement, advocating the Direct Method of language
teaching, as laid out in his 1899 book.
According to Jones (1941), Passys greatness lay not
chiefly in his phonetic work, but in the saintly way
he lived his life according to his primitive Christian
ideals, after his conversion to Protestantism in 1878,
and according to his Christian socialist ideals from
1897. He formed a spartan agricultural cooperative,
Lie fra (an acronym of the French ideals of liberty,
equality, and fraternity), for working-class men near
Fontette. This, along with his ideals, evangelization,
and social work, is detailed in his 19301932 autobiography and in his novels Au bois dormant and Apre`s
le re ve. He was the founder of the Socie te des Volontaires E vangelistes and the Union des Socialistes
Chre tiens. Passy died in Bourg-la-Reine in 1940.
See also: International Phonetic Association; Bell, Alexan-
Bibliography
Collins B & Mees I M (1999). The real Professor Higgins:
the life and career of Daniel Jones. Berlin: Mouton de
Gruyter.
Jones D (1941). Paul Passy. In Le matre phone tique 3rd
series 75, 3039. [Reprinted (in original phonetic transcription) in Sebeok T A (ed.) (1966) Portraits of linguists: a biographical source book for the history of
Western linguistics, 17461963 (vol. 2). Bloomington:
Indiana University Press. 139147.]
Jugnet L (1929). Paul Passy: un apo tre. Laon: Imprimerie
des Tablettes de lAisne.
Michaelis H & Passy P (1897). Dictionnaire phone tique de
la langue franc aise. Hanover: Carl Meyer.
Passy P (1887). Les sons du franc ais: leur formation, leur
combinaison. Paris: Firmin Didot. [Translated (1907)
into English by Davory D L & Jones D as The sounds
of the French language, their formation, combination and
representation. Oxford: Clarendon.]
Passy P (1891). E tude sur les changements phone tiques et
leurs caracte`res ge ne raux. Paris: Firmin-Didier.
Passy P (1899). De la me thode directe dans lenseignement
des langues vivantes. Paris: Colin.
Passy P (1914). A French phonetic reader. London: University of London Press.
Passy P (1929). La phone tique et ses applications. Association Phone tique Internationale.
Passy P (19301932). Souvenirs dun socialiste chre tien
(2 vols). Issy-les-Moulineaux: Editions Je sers.
most important
(5th century B.C.) is considered the
grammatical text. The major commentaries on the
Astadhyay are the Varttikas (Explanatory Rules) of
tyayana (3rd century B.C.) and the Mahabhasya of
Ka
classified into 85 ahnikas (day sessions), is considered the second most important grammatical text
after Paninis Astadhyay. In it, Patanjali takes up
Katyayanas
varttikas,
explains them with examples,
and in many cases defends Panini. Moreover, he also
carries out in great measure Katyayanas work by
examining sutras on which there are no varttikas.
To see how Patanjali explains sutras and varttikas,
consider the following:
Panini 2.3.70 (sutra dealing with the sixth (genitive)
case ending)
akenorbhavisydadhmarnyoh. (The gen is
and object
not added
itive in the sense of agent
when the words ending in the suffix) aka or in (is
used) denoting the sense of bhavisyat (future (ac
tion)) or adhamarnya (indebtedness).
Katyayanas clarification varttikas: (a) akasya bhavisyati (the mention) of (the suffix) aka (must be
taken)
in the sense of bhavisyat (future (action));
Patanjalis
explanation bhasyas on Katyayanas
bhavisyatti vaktaclarification varttikas: (a) akasya
ment should be made that (the mention) of (the suffix) in in the sense of adhamarnya (indebtedness)
also. (For example), one who is in indebtedness to
pay one hundred, one who is in indebtedness to pay
one thousand, one who intends to go to the village.
As can be seen, the Mahabhasya is written in a
which takes the
lively, simple, and animated style,
form of an actual conversation. There are occasions
when Patanjali gives a dynamic picture of the mode of
discussion current in his time. Proverbial expressions
and references to the matters of daily life are often
introduced to enliven the discussions, and these give
valuable hints to the conditions of life and thought at
the time of Patanjali (Keith, 1948: 428). For example,
consider Patanjalis comment on the Sanskrit suffix
ka, which when added to a name denotes an image of
that name, e.g., hastika (the image of an elephant),
asvaka (the image of a horse). In sutra 4.3.99, however, Panini says jvikarthe capanye: (The suffix ka is
when (the image) is used
dropped)
to secure a livelihood and it is not for sale. Thus, when a magician
uses the image of an elephant during his show, the
word for the image is hastin, but in a toy store where
it is for sale, it is hastika, as in hastikan vikrnte (He
sells the images of elephant.) (Vasu, 1962: Vol. I:
975). Commenting on this sutra, Patanjali says: apanya ity ucyate tatredam na sidhyati sivah skando
akha iti. kim karnam? mauryair hiranya
rthibhir
vis
samprati
pujartha s tasu bhavisyati. (The word) apa
nya (not to be vendible) is mentioned (in the sutra).
this doctrine the forms sivah, skando, visakha are
On
incorrect. Why is that? Because the Mauryas, in their
greed for gold, used as means the images of gods (i.e.,
they sold the images of these gods and called them
sivah, skando, visakha, not the expected sivaka, skan visakhaka). It is granted that the rule for
daka,
dropping the suffix ka does not apply to those images
of the (greedy) Mauryas; still, as regards images now
used for purposes of worship. (i.e., for livelihood only
and not for sale) it does apply (Agnihotri, 1963:
552).
The history of grammatical thought in India shows
that, with the exception of Bhartrhari (7th century
itself became unintelligible and evidently needed explanation. Bhartrhari, who in his Va kya-Padya
(That which enlightens the Maha bha sya) com had been
mented on the Maha bha sya, which in turn
itself.
See also: Bhartrhari; Dikhsita, Bhattoji (ca. 17th Century
A.D.); Katyayana (3rd Century B.C.); Nagojibhatta (d. 1755);
Panini.
Bibliography
Abhyankar K V (ed.) (1960). Paribha sendus ekhara of
Na ges a. English and notes Kielhorn F (trans.). Poona:
Bhandarkar Oriental Research Institute.
Abhyankar K V & Limaye V P (eds.) (1965). Va kya-Padya
of Bhartrhari. Poona: University of Poona.
Agnihotri P (1963). Patan jalika lna Bha rata. Patna:
Bihar-Rashtrabhasha-Parishada.
Agrawal V S (1963). India as known to Panini. Varanasi:
Prithivi Prakashan.
Allen W S (1953). Phonetics in ancient India. London:
Oxford University Press.
Belvalkar S K (1976). An account of different existing
systems of Sanskrit grammar. (Reprint). Delhi: Bhartiya
Vidya Prakashan.
Bhartrhari (1922). Subhasita Tris ati. Bombay: Nirnaya
Press.
Sagar
Bloomfield L (1927). On Some Rules of Pa nini. Journal of
the American Oriental Society 47, 6170.
Bloomfield L (1929). Review of Leibich. Language 5,
267275.
Bloomfield L (1931). Language. New York: Henry Holt.
itself became unintelligible and evidently needed explanation. Bhartrhari, who in his Vakya-Padya
itself.
See also: Bhartrhari; Dikhsita, Bhattoji (ca. 17th Century
A.D.); Katyayana (3rd Century B.C.); Nagojibhatta (d. 1755);
Panini.
Bibliography
Abhyankar K V (ed.) (1960). Paribhasendusekhara of
Nagesa. English and notes Kielhorn F (trans.). Poona:
Bhandarkar Oriental Research Institute.
Abhyankar K V & Limaye V P (eds.) (1965). Vakya-Padya
of Bhartrhari. Poona: University of Poona.
Agnihotri P (1963). Patanjalikalna Bharata. Patna:
Bihar-Rashtrabhasha-Parishada.
Agrawal V S (1963). India as known to Panini. Varanasi:
Prithivi Prakashan.
Allen W S (1953). Phonetics in ancient India. London:
Oxford University Press.
Belvalkar S K (1976). An account of different existing
systems of Sanskrit grammar. (Reprint). Delhi: Bhartiya
Vidya Prakashan.
Bhartrhari (1922). Subhasita Trisati. Bombay: Nirnaya
Press.
Sagar
Bloomfield L (1927). On Some Rules of Panini. Journal of
the American Oriental Society 47, 6170.
Bloomfield L (1929). Review of Leibich. Language 5,
267275.
Bloomfield L (1931). Language. New York: Henry Holt.
(18201885); Leskien, August (18401916); Saussure, Ferdinand (-Mongin) de (18571913); Steinthal, Heymann
(18231899); Zarncke, Friedrich (18251891).
Bibliography
Antal L (1985). Some comments on the relationship between [Hermann] Paul and Saussure. Cahiers Ferdinand
de Saussure 39, 121130.
Cherubim D (1973). Hermann Paul und die moderne
Linguistik: Zur Studienausgabe von H. Pauls Prinzipien
der Sprachgeschichte. Zeitschrift fu r Dialektologie und
Linguistik 40, 310322.
Koerner E F K (1972). Hermann Paul and synchronic
linguistics. Lingua 29, 274307.
Paul H (1880). Principien der Sprachgeschichte. Halle:
Max Niemeyer, repr. Max Niemeyer, Tu bingen, 1970;
Strong H A (English trans.), Principles of the History of
Language, Sonnenschein, London, 1888, Macmillan,
New York, 1889; new revised edn. London: Longmans,
Green, 1890; repr., McGrath College Park, MD, 1970.
Paul H (19161920). Deutsche Grammatik (5 vols). Halle:
Max Niemeyer.
Paul H (1922). Mein Leben. Beitra ge zur Geschichte der
deutschen Sprache und Literatur 46, 495500.
Reis M (1978). Hermann Paul. Beitrage zur Geschichte
der deutschen Sprache und Literatur 100, 159204.
Streitberg W (1922). Hermann Paul. Indogermanisches
Jahrbuch 9, 280285.
Strong H A, Logeman W S & Wheeler B I (1891).
Introduction to the Study of the History of Language.
London: Longmans, Green.
(18201885); Leskien, August (18401916); Saussure, Ferdinand (-Mongin) de (18571913); Steinthal, Heymann
(18231899); Zarncke, Friedrich (18251891).
Bibliography
Antal L (1985). Some comments on the relationship between [Hermann] Paul and Saussure. Cahiers Ferdinand
de Saussure 39, 121130.
Cherubim D (1973). Hermann Paul und die moderne
Linguistik: Zur Studienausgabe von H. Pauls Prinzipien
der Sprachgeschichte. Zeitschrift fur Dialektologie und
Linguistik 40, 310322.
Koerner E F K (1972). Hermann Paul and synchronic
linguistics. Lingua 29, 274307.
Paul H (1880). Principien der Sprachgeschichte. Halle:
Max Niemeyer, repr. Max Niemeyer, Tubingen, 1970;
Strong H A (English trans.), Principles of the History of
Language, Sonnenschein, London, 1888, Macmillan,
New York, 1889; new revised edn. London: Longmans,
Green, 1890; repr., McGrath College Park, MD, 1970.
Paul H (19161920). Deutsche Grammatik (5 vols). Halle:
Max Niemeyer.
Paul H (1922). Mein Leben. Beitrage zur Geschichte der
deutschen Sprache und Literatur 46, 495500.
Reis M (1978). Hermann Paul. Beitrage zur Geschichte
der deutschen Sprache und Literatur 100, 159204.
Streitberg W (1922). Hermann Paul. Indogermanisches
Jahrbuch 9, 280285.
Strong H A, Logeman W S & Wheeler B I (1891).
Introduction to the Study of the History of Language.
London: Longmans, Green.
Bibliography
Paulston C B (1980). Bilingual education. Theories and
issues. Rowley, MA: Newbury House.
Paulston C B (1994). Linguistic minorities in multilingual
settings. Implications for language policies. Studies in
Bilingualism 4. Amsterdam/Philadelphia: John Benjamins.
Paulston C B (ed.) (1998). International handbook of bilingualism and bilingual education. New York: Greenwood.
Paulston C B & Peckham D (eds.) (1998). Linguistic minorities in Central and Eastern Europe. Clevedon, UK:
Multilingual Matters.
Bibliography
Paulston C B (1980). Bilingual education. Theories and
issues. Rowley, MA: Newbury House.
Paulston C B (1994). Linguistic minorities in multilingual
settings. Implications for language policies. Studies in
Bilingualism 4. Amsterdam/Philadelphia: John Benjamins.
Paulston C B (ed.) (1998). International handbook of bilingualism and bilingual education. New York: Greenwood.
Paulston C B & Peckham D (eds.) (1998). Linguistic minorities in Central and Eastern Europe. Clevedon, UK:
Multilingual Matters.
Timing
Delays are a matter of timing. When people talk, they
recognize that there is a conventional way of producing each sentence in contextits ideal delivery.
An ideal delivery is characteristically a single action
with no suspensionsno silent pauses, no fillers, no
repeats, no self-corrections, no delays except for
those required by the syntax of the sentence. For
Alans utterance in (1), the ideal delivery would be
but at the same time I did accuse them of having
misled us on April the twenty-third. That makes the
ideal delivery a model against which speakers can
judge their actual delivery. It defines what it is to be
a delaya temporary suspension of the action of
producing the conventional form of that utterance.
Reactive delays arise when speakers are unable to
produce an ideal delivery. Rhetorical delays arise
when speakers choose to suspend an ideal delivery
for rhetorical purposes.
In conversation, there are also conventions about
how to go from one utterance to the next. In many
styles of conversation, speakers are expected to move
smoothly, without undue gaps, both between units
within turns and between turns (Sacks et al., 1974).
In (1), for example, Alans partner Ben is allowed to
start speaking once Alan has completed the units
ending misled us and twenty-third, and if Ben
does start first at either point, he gets the next turn
from then on. This practice, applied generally, tends
to minimize gaps both between units within turns and
between turns.
The conventional standards of fluency, however,
are almost impossible to live up to. Speakers cannot
produce an expression until they have formulated it,
and it may take longer to formulate an expression
than time allows. The result is often unwanted delays.
Also, people in conversation owe their partners an
account for any extra time they take. If they delay,
they may want to say when and why they are doing
so. In (2), Duncan has been just asked about recent
books he has read:
(2) Duncan Ive u:m recently read u:m . oh, . Lord
of the Flies (3.5a.110)
only by monitoring their progress that they can anticipate when to delay and what comments to add.
Speakers must also monitor their partners. If their
partners arent attending or understanding, or are
starting to speak themselves, speakers may need to
stop to deal with those problems as well.
Formulating Delays
Delays dont just happen. As parts of the speech
stream, they need to be formulated with as much
care as the speech itself. The act of delaying has
three phases: (1) the suspension of fluent speech;
(2) a hiatus in which speakers may say and do other
things; and (3) the resumption of fluent speech
(Clark, 1996). In (2), fluent speech was suspended
at the end of read and resumed at the beginning of
Lord. The hiatus, the interval in between, contained
u:m . oh, .
Suspending Speech
Julia apparently had trouble describing the building, so she delayed while choosing monastery and
again while choosing very Gothic monastery. She
signaled her first point of suspension to begin at
the end of the by pronouncing it thi:y. In one
large sample of conversation, speakers suspended
their speech 81% of the time after thiy, but only
7% of the time for thuh. So when thuh is the
appropriate pronunciation for the, speakers can
signal a suspension of speech by using thiy instead.
They can do the same by using the pronunciations
of a and to that rhyme with day and flew (Fox Tree
and Clark, 1997).
Hiatuses
Hiatuses end when speakers resume speaking fluently. Speakers have four main options in resuming,
once again as documented in spontaneous speech.
1. Continuations. Speakers most often continue
a sentence where they left off, picking up on the
next word with the intonation of the phrase they
suspended. This is illustrated in examples 2 and 5.
2. Restarts. Speakers also often return to the beginning of the current constituent and restart it, as
Julia did in (4) with the, its thi:y . thuh monastery. Evidence shows that speakers are more likely to restart a constituent the more disruptive the
hiatus has beenthe more commentary or time
it contained (Clark and Wasow, 1998). So, even
though their original attempt at a constituent may
have been interrupted, their ultimate delivery of
the constituent is fluent, befitting an ideal delivery.
There is good evidence that speakers prefer to
produce constituents with a continuous delivery
(Clark and Wasow, 1998; Levelt, 1983).
3. Replacements. Still other times, when speakers
resume speaking, they replace earlier material
in order to correct an error, as in Sunday . the
twenty-fifth, - sorry twenty-fourth (211a.173).
4. Fresh starts. Speakers sometimes resume speaking
with a fresh start on an entirely new sentence,
as in (3).
Why Delay?
If speakers delay in order to deal with problems in
production, what types of problems are they? One of
the earliest proposals was that delays come at points
of unpredictability in production (Goldman-Eisler,
1968; Maclay and Osgood, 1959). Unpredictability
was measured by deleting words from transcripts of
spontaneous speech and counting the number of
guesses it took other people to identify the deleted
words. The more guesses it took for a word, the
less predictable the word was in context. Indeed,
speakers tended to pause longer and more often just
before the words that were later shown to be less
predictable.
According to another early proposal, speakers
should also be delayed in formulating major units of
their utterances. One such unit is the intonation unit
(Boomer, 1965). These units are clauses or other
phrases that are pronounced under a single intonation
contour. Intonation units, in this proposal, should
have many delays near their beginnings, where most
planning is done, and fewer delays later on. Indeed,
pauses, fillers, and repeats have been shown to be
common before or after the first word of intonation
units and rare later on (Chafe, 1979, 1980a; Clark
and Fox Tree, 2002). Likewise, within intonation
units, there tend to be more delays before long
phrases than before short ones (Clark and Wasow,
1998; Ford and Holmes, 1978; Holmes, 1988).
Narratives consist of series of intonation units, as
in this excerpt from a retelling of a silent film (each
line is an intonation unit) (Chafe, 1980b):
(6) (1.15) And (.1) then a boy comes by,
(.1) on a bicycle,
the man is in the tree,
(.9) and the boy gets off the bicycle,
and .. looks at the man,
and then (.9) uh looks at the bushels,
and he .. starts to just take a few,
and then he decides to take the whole bushel.
Summary
Delays are common in spontaneous speech. Reactive
delays are almost inevitable as speakers try to remain
fluent while deciding on, planning, and formulating
what to say. Rhetorical delays, in contrast, are part
and parcel of what speakers are trying to say. The
remarkable point is that delays themselves are
planned. Speakers must formulate each suspension,
each hiatus, and each resumption with the same care
that they formulate their speech.
See also: Dialogue and Interaction; Psycholinguistics:
Overview; Speech Errors: Psycholinguistic Approach;
Spoken Language Production: Psycholinguistic Approach.
Bibliography
Blackmer E R & Mitton J L (1991). Theories of monitoring and the timing of repairs in spontaneous speech.
Cognition 39, 173194.
Boomer D S (1965). Hesitation and grammatical encoding. Language and Speech 8, 148158.
Chafe W (1979). The flow of thought and the flow
of language. In Givon T (ed.) Syntax and semantics 12:
discourse and syntax. New York: Academic Press.
159181.
Chafe W (1980a). The deployment of consciousness in the
production of a narrative. In Chafe W (ed.) The pear
stories. Norwood NJ: Ablex. 955.
Chafe W (ed.) (1980b). The pear stories. Norwood, NJ:
Ablex.
Summary
Delays are common in spontaneous speech. Reactive
delays are almost inevitable as speakers try to remain
fluent while deciding on, planning, and formulating
what to say. Rhetorical delays, in contrast, are part
and parcel of what speakers are trying to say. The
remarkable point is that delays themselves are
planned. Speakers must formulate each suspension,
each hiatus, and each resumption with the same care
that they formulate their speech.
See also: Dialogue and Interaction; Psycholinguistics:
Overview; Speech Errors: Psycholinguistic Approach;
Spoken Language Production: Psycholinguistic Approach.
Bibliography
Blackmer E R & Mitton J L (1991). Theories of monitoring and the timing of repairs in spontaneous speech.
Cognition 39, 173194.
Boomer D S (1965). Hesitation and grammatical encoding. Language and Speech 8, 148158.
Chafe W (1979). The flow of thought and the flow
of language. In Givon T (ed.) Syntax and semantics 12:
discourse and syntax. New York: Academic Press.
159181.
Chafe W (1980a). The deployment of consciousness in the
production of a narrative. In Chafe W (ed.) The pear
stories. Norwood NJ: Ablex. 955.
Chafe W (ed.) (1980b). The pear stories. Norwood, NJ:
Ablex.
Types of Grammars
There are several types of grammars in existence:
linguistic grammars (descriptions), reference grammars, pedagogical grammars, and language courses
(the last with grammar components). Linguistic grammars or descriptions are primarily for testing linguistic
theories; two examples of such grammars are Jolly
(1991), which presents a role and reference grammar
of English, and Haegeman and Gue ron (1999), which
is a transformational, or generative, grammar of
English. Since the purpose of such grammars is well
defined in terms of theoretical goals, they are usually
limited in scope (e.g., Ginzburg and Sag, 2000), or
when they attempt to be comprehensive, they tend to
propose modifications and extensions of theories
they are based on. A grammar driven by linguistic
theory may also illustrate how the theory works.
A good example of a descriptive grammar driven by
linguistic theory is Halliday (1985), which illustrates
how the systemic model is effective in describing
English.
Reference grammars are usually theory neutral,
i.e., they derive insights from several theoretically
oriented descriptions and present information in traditional grammatical terms or in terms that are
explained explicitly so that a wide range of users
may find the information useful. Two good examples
of reference grammars of English are Quirk et al.
(1985) and Huddleston and Pullum (2002).
Language courses present language material with a
view to facilitating learning; the emphasis is on
making the units interesting and useful for communication via plentiful, but not overwhelming, structural, lexical, and sociocultural content. Grammatical
points are explained, but they are only one of the several concerns of the course writers. Since it is tied to
the lessons, the grammatical information is fragmentary and not systematic as in a reference grammar.
Examples of language courses with grammatical
notes are Rutherford (1975) for English and Kachru
and Pandharipande (1983) for Hindi.
One distinctive characteristic of pedagogical grammars is that they facilitate awareness of the relatedness of grammatical structures to speaker/writer
meanings and intentions. The primary concern of linguistic grammatical descriptions is linguistic theory;
that of reference grammars is presenting grammatical
topics in a systematic way in order to focus on the
grammaticalness of structures. Issues of sociocultural appropriateness are not central to either type.
The following topics may be said to be the essential properties of pedagogical grammars; they are
illustrated with material drawn from a number of
Western and non-Western languages.
Relating Use (Function) and Usage (Grammatical
Rules)
(indefinite article;
nonspecific, generic)
(12b)
(13a) The
(13b)
I saw a girl
jogging in
the snow
(indefinite article;
specific, unidentified)
(definite article,
nonspecific,
generic)
(definite article,
specific,
identified)
kha
eat
ljiye
take HON IMP
(15) kha na
meal
kha
eat
ljiyega
take HON IMP FUT
(16) kha na
meal
kha
eat
le
take OPT
(17) kha na
meal
kha
eat
liya
take PERF
ja e
go OPT
Hindi makes a distinction between intimate, familiar, and honorific forms of second person pronouns
and imperative forms of verbs. The above are all
honorific forms. As the glosses suggest, each one
represents a different grammatical construction. In a
pedagogical grammar, however, they belong together
with explanations of contexts in which each one is
appropriate: (14) is used for intimate or very familiar
addressees one respects, e.g., ones parents, older siblings, close relatives; (15) is used when the context
demands that the addressee be given the impression
that he or she has a choice not to eat immediately but
whenever it is convenient the distance in time suggests a higher degree of politeness; (16) is used in very
formal situations since the verb is not marked with
the imperative ending but rather the optative suffix,
which signals a suggestion; and (17), a passive verb
with an optative suffix, is used especially in the
eastern Hindi-speaking area to suggest the highest
degree of politeness.
Interactional Norms of Speech Communities
The grammatical generalization in terms of attribution does not help an English-speaking learner of
French attempting to figure out where to use what
seems like an article [les, le, in (18) and (19), respectively] and where not to use it [e.g., in (20)], especially
if the learner wants to participate in interactions as a
speaker, as opposed to decoding what has been said,
as a listener. It also does not lead to the learners
internalizing the nature of the attributive sentences
in French. A contrastive account of English and
French article use such as the following may be helpful: a predicate nominal complement of a linking verb
in French typically does not take an indefinite
article as in English; on the other hand, the name of
Conclusion
In the teaching of languages and grammars, the emphasis has always been on the target language; some
approaches prohibit use of learners first languages in
the classroom. Since pedagogical grammars are meant
for serving specific purposes, comparison with first or
any other familiar languages in context need not be
avoided. In fact, contrasting the target language properties with the characteristics of other languages in
the learners repertoire may be a very effective tool
in helping learners internalize the target language
grammars.
See also: Communicative Competence; Communicative
Language Teaching; Hindi; Honorifics; IndoAryan Languages; Interlanguage; Language Change and Language
Contact; Language Education: Grammar; Speech Community.
Bibliography
Algeo J (1988). British and American grammatical
differences. International Journal of Lexicography 1(1),
131.
Berns M (1990). Contexts of competence: social and cultural considerations in communicative language teaching.
New York: Plenum Press.
Celce-Murcia M (1995). Learning and teaching L1 and
L2 grammar: many differences but also some similarities. Revue de IACLA 3D journal of the CAAL 17(2),
4755.
Ciliberti A (1984). Towards a definition of pedagogical
grammars. Studi Italiani di Linguistica Teorica ed
Applicata 13(1), 2344.
Coulmas F (ed.) (1981). Conversational routine: explorations in standardized communication situations and
prepatterned speech The Hague: Mouton.
Eades D (1982). You gotta know how to talk . . . : information seeking in South-East Queensland aboriginal
society. Australian Journal of Linguistics 1982 2(1),
6182.
Ellis R (2002). The place of grammar instruction in the
second/foreign curriculum. In Hinkel E & Fotos S (eds.)
New perspectives on grammar teaching in second
language classroom. Mahwa, NJ: Erlbaum. 1734.
Eskey D E (1983). Meanwhile back in the real world . . . :
accuracy and fluency in second language teaching.
TESOL Quarterly 17(2), 315323.
Ginzburg J & Sag I (2000). Interrogative investigations:
the form, meaning, and use of English interrogatives.
Stanford: CSLI Publications. [distributed by University
of Chicago Press.]
Haegeman L M V & Gue ron J (1999). English grammar: a
generative perspective. Oxford: Blackwell.
Halliday M A K (1985). An introduction to functional
grammar. London: Edward Arnold.
Huddleston R & Pullum G K (2002). The Cambridge grammar of the English language. Cambridge: Cambridge
University Press.
Ilson, Robert (1995) A(n)-dropping. English Today 11(1),
4243.
Jolly J (1991). Prepositional analysis within the framework
of role and reference grammar. New York: Peter Lang.
Kachru Y & Pandharipande R (1983). maadhyamik hindii:
intermediate Hindi. Delhi: Motilal Banarsidass. [latest
reprint: 2003.]
Krashen S (1981). Second language acquisition and second
language learning. Oxford: Oxford University Press.
Labov W (1998). Co-existent systems in African-American
vernacular English. In Mufwene S S, Rickford J R,
Bailey G et al. (eds.) African-American English: structure, history and use. New York: Routledge. 110153.
Lightbown P M (2000). Classroom SLA research and
second language teaching. Applied Linguistics 21(4),
431462.
Martin S E (1964). Speech levels in Japan and Korea. In
Hymes D (ed.) Language in culture and society. New
York: Harper and Row. 407415.
McArthur T (1995). The absent indefinite article. In
response to Robert Ilson (1995). English Today 11(1),
4344.
McDonald E (1999). Teaching grammar through text: an
integrated model for a pedagogical grammar of Chinese.
Journal of Chinese Language Teachers Association 34(2),
91120.
Nassaji H & Fotos S (2004). Current developments in
research and teaching of grammar. Annual Review of
Applied Linguistics 24, 126145.
Norris J M & Ortega L (2000). Effectiveness of L2 instruction: a research synthesis and quantitative meta-analysis.
Language Learning 50, 417428.
Norris J M & Ortega L (2001). Does type of instruction
make a difference? Substantive findings from a metaanalytic review. In Ellis R (ed.) Form-focused instruction
and second language learning. Malden, MA: Blackwell.
157213.
Quirk R, Greenbaum S, Leech G et al. (1985). A comprehensive grammar of the English language. London:
Longman.
Rutherford W E (1975). Modern English. (2 vols; 2nd edn.).
New York: Harcourt Brace Jovanovich.
Savignon S J (ed.) (2002). Interpreting communicative
language teaching: contexts and concerns in teacher
education. New Haven: Yale University Press.
Schmidt R W (2001). Attention. In Robinson P (ed.)
Cognition and second language instruction. Cambridge:
Cambridge University Press. 332.
Skehan P (1998). A cognitive approach to language
learning. Oxford: Oxford University Press.
Teng S (1997). Towards a pedagogical grammar of
Chinese. Journal of the Chinese Language Teachers
Association 32(2), 2939.
VanPatten B (1988). How juries get hung: problems with
the evidence for a focus on form in teaching. Language
Learning 38(2), 243260.
Bibliography
Bentzen R (ed.) (1994). Ung sprogforsker pa rejse. Breve fra
og til Holger Pedersen 18921896. Copenhagen: Det
Kongelige Bibliotek.
Hendriksen H (1937). Bibliographie des publications de
Holger Pedersen. In Hjelmslev L et al. (eds.) Melanges
linguistiques offerts a` M. Holger Pedersen. Aarhus:
Universitetsforlaget. IXXXVII.
Henriksen C (1996). Pedersen, Holger. In Lexicon Grammaticorum: whos who in the history of world linguistics.
Tubingen: Niemeyer. 710.
Henriksen P A (1983). Holger Pedersen a comprehensive
bibliography of his writings. In Pedersen H A glance at
the history of linguistics with particular regard to the
historical study of phonology. Henriksen C C (trans.)
and Koerner K (ed.). Amsterdam: BenjaminsXXX.
Knudtzon J A (1902). Die zwei Arzawa-Briefe, die altesten Urkunden in indogermanischer Sprache. Leipzig:
Hinrichs.
Pedersen H (1895). Albanesische Texte mit Glossar. Leipzig:
Teubner.
Pedersen H (19091913). Vergleichende Grammatik der
keltischen Sprachen (2 vols.). Gottingen: Vandenhoeck
& Ruprecht. [Rept. 1976.]
Pedersen H (1931). Linguistic science in the 19th century:
the discovery of language. Cambridge: Harvard University Press. [Danish original 1924.]
Pedersen H (1938). Hittitisch und die anderen indoeuropaischen Sprachen. Kbenhavn: Munksgaard.
Pedersen H (1941). Tocharisch vom Gesichtspunkt der
indoeuropaschen Sprachvergleichung. Kbenhavn:
Munksgaard.
Pedersen H (1945). Lykisch und Hittitisch. Kbenhavn:
Munksgaard.
Schmitt R (ed.) (1982). Pedersen, H., Kleine Schriften zum
Armenischen. Hildesheim and New York: Olms.
Pehuenche
See: Mapudungan.
His Work
Life
Charles Sanders Peirce (pronounced purse) was born
on September 10, 1839, in Cambridge, Massachusetts.
He was regarded as a child prodigy in science
and philosophy. His father, Benjamin Peirce, was a
renowned mathematician at Harvard University who
liked to stimulate his sons thinking by presenting
intricate and original problems. This unique didactic
atmosphere may have been a stimulus for Peirces
development as an original thinker. However, despite
a promising youth and a brilliant mind, Peirce never
achieved a tenured academic position.
Peirce graduated from Harvard in 1859 and received
his bachelors degree in chemistry from Lawrence Scientific School in 1863. That same year he married his
first wife, Harriet Melusina Fay. For the most part of
his life i.e., from 1859 until the end of 1891 he was
an employee with the U.S. Coast and Geodic Survey (of
which his father was a cofounder), where he performed
gravity research and pendulum experiments. His
work brought him to Europe on five occasions, each
time for a stay of several months. From 1879 until
1884, Peirce held a parallel job at Johns Hopkins University, teaching logic at the department of mathematics. His early resignation is said to be due to his personal
problems (he got divorced and later remarried).
After his forced resignation from the U.S. Coast
and Geodic Survey in 1891, Peirce was never again
able to obtain a steady job, although he was invited
for short-term lecturing posts at various institutions
such as Harvard (1865, 186970, 1903, 1907) and
the Lowell Institute in Boston (1866, 1892, 1903).
Most of his writing after 1891 was done for pay.
Peirce reviewed and translated books and contributed
to several dictionaries and encyclopedias. His articles,
in which he worked out his theories, often remained
unfinished or were rejected by editors. He also did
some consultancy work as a chemist. During the last
years of his life, despite inheriting money and some
possessions from his mother and aunt, he often had
to rely on the financial support of his friends, e.g.,
Another triadic distinction is applied to his classification of signs. In its simplest form, Peirces classification
distinguishes ten classes of signs based on three trichotomies that are themselves derived from his three
universal categories. In later work, however, Peirce
argued, albeit without offering a satisfying analysis,
that there are no less than sixty-six classes based on
ten trichotomies.
The first trichotomy is based on the character of the
sign itself. Qualisigns are qualities (e.g., hardness)
which may act as a sign; signs which are actual events
or things (such as a in the previous sentence) are sinsigns, and a regularity (or law) that is a sign is
termed a legisign (e.g., the determiner a as a linguistic
item).
The second trichotomy is based on the relationship
between the sign and its object. An icon is a sign that
refers to its object merely because of the features it
possesses. It is important to note that sheer similarity
between sign and object is not what makes a sign an
icon; rather, iconicity is based on the fact that the two
are interpreted as similar. The second type of sign in
relation to the object is called index. An index is a
sign that is affected by its object, in other words there
is a real physical relationship between the two. One
popular example is smoke as a sign of fire. Fire produces smoke, so there is a direct (causal) relationship.
Finally, a symbol is a sign that refers to its object only
because there is an interpretant who links the sign to
the object, e.g., words and other conventional signs.
This second trichotomy is the most interesting
(and most widely known) one from a linguistic
point of view as it addresses the time-honored topic
of the arbitrariness of the linguistic sign and whether
there is iconicity involved in language (see Saussure,
Ferdinand (-Mongin) de (18571913)). The linguist
R. Jakobson was the first to use Peirces terminology
to address this topic in his influential article Quest
for the essence of language (1966). Since the 1980s a
growing number of linguists (mainly in the wake of
the work of John Haiman) have argued against arbitrariness in language and in favor of the pervasiveness
of iconicity.
The third trichotomy is concerned with how the
sign is interpreted and corresponds to the older distinction between term, proposition, and argument.
A rheme is understood to represent its object merely
in its possibilities or features and is neither true nor
false (see Aristotle and Linguistics). A dicent sign or
dicisign represents its object in respect to actual existence and as such it is either true or false. Finally, an
argument signifies a law leading from a set of premises to a conclusion.
One of the hallmarks of Peirces work is his disappointment with the results of philosophy compared
with those of the natural sciences. According to Peirce,
the success of science is due to its methodology; he
argued that philosophy should adopt the same methods, including the use of a well-defined nomenclature.
Pragmatism (which Pierce renamed pragmaticism in
his later writings) may be regarded as an attempt to
establish philosophy as a science, i.e., a method of
thinking through which the impact of concepts on
mans conduct can be assessed. Thus, pragmatism is
ultimately a method to ascertain the true meanings,
understood as possible pragmatic values, of concepts.
Later pragmatists such as William James and John
Dewey extended pragmaticism to other issues as well
(e.g., the problem of truth).
Peirce argued that abduction is fundamental in
scientific as well as everyday reasoning, being the
only type of inference that leads to genuinely new
ideas and theories something that induction (i.e.,
reasoning from some given facts to a general law) and
deduction (reasoning from a general idea to a particular case) never do. Abduction is the creative process
of forming and provisionally accepting an explanatory hypothesis for the purpose of testing it. However,
since abduction can only suggest what may be the
case, it can ultimately never be more than ordinary
guesswork.
See also: Aristotle and Linguistics; Jakobson, Roman
(18961982); Kant, Immanuel (17241804); Saussure, Ferdinand (-Mongin) de (18571913).
Bibliography
Brent J (1993). Charles Sanders Peirce: a life. Bloomington:
Indiana University Press.
Jakobson R (1971). Quest for the essence of language. In
Jakobson R (ed.) Selected writings. II. Word and language. Mouton: The Hague. 345359. [Originally published 1966.]
Peirce C S (19311958). Collected papers of Charles Sanders
Peirce. Hartshorne C & Weiss P (eds.) (vols IVI) and
Burks A (ed.) (vols VIIVIII). Cambridge: Harvard
University Press.
Peirce C S (1982). Writings of Charles Sanders Peirce:
A chronological edition (vols 16). Bloomington:
Indiana University Press.
Peirce C S (19921998). The essential Peirce: selected philosophical writings, vol. 1. Houser N and Kloesel C (eds.)
and vol. 2, Peirce Edition Project (ed.). Bloomington:
Indiana University Press.
Transactions of the Charles S. Peirce society. A quarterly
journal in American philosophy. Buffalo, NY: Philosophy
department, SUNY Buffalo. [Published since 1965.]
Perceptual Dialectology
D R Preston, Michigan State University, East Lansing,
MI, USA
2006 Elsevier Ltd. All rights reserved.
question (4) were treated separately. When respondents performed similarly in stating where differences were, they were grouped into a subjective
speech community. Sibata and his colleague W. A.
Grootaers (1959) claimed that such boundaries
were of little interest to linguists, since they do not
generally correspond to production boundaries.
Y. Mase (1964a) asked respondents to indicate
surrounding areas that sounded the same or different,
and provided maps based on both, since the Alpine
Japanese in his survey were willing to name both
kinds of sites. He first mapped responses to two questions: (1) which sites sound the same? and (2) which
sites sound a little different?; the perceptual areas
made up of the responses are based on reciprocal
perceptions of similarity, on similar perceptions of
the first degree of difference, and on the perception
by surrounding areas of their similarity to one another. Mase (1964b) also compared his perceptual
boundaries to traditional isoglosses and found a good
correspondence. On the other hand, he and Nomoto
(1963) found school districts, rather than feudal and
other political administrative zones, areas that dominated the boundaries of the Itoigawa research, to be
similar to perceptual areas. Weijnen (1968) criticized
the Itoigawa research by noting that respondents
were asked if there were differences rather than similarities, and praised Mase, but since Mase nowhere
based maps exclusively on similarity, Weijnen might
not have approved if he had had access to the original
(Japanese) versions of Mases work.
Motivated by a desire to explore folk knowledge
for its own value rather than for only the perception
production correlation, Preston (1996) asked U.S.
respondents to rank areas on a scale of one to four
(1 same, 2 a little different, 3 different,
4 unintelligibly different). When many U.S. respondents evaluate differences, they perceive a rather large
local area of similarity, but ratings of and from the
U.S. South are of greatest interest. Among northern
raters, a large South emerges as a territory rated 3
(different), often with a core South rated 4 (unintelligibly different). Southern raters also rank a fairly
sizeable New England (i.e., northeastern) area as 4.
Evaluations of the degree of difference in this style
and with slight variations have been done in southeastern Michigan and southern Indiana, the U.S.
South, California, and Oregon. Sites outside the
United States include The Netherlands, Norway,
Canada, Spain, France, Italy, Turkey, Switzerland,
and Germany (e.g., Preston, 1999; Long and
Preston, 2002). In addition to the reporting of
means scores, some of these studies have employed
Figure 1 Speech regions for southeastern Michigan respondents. Key to map regions: 1, South; 2, North; 3, Northeast; 4, Southwest;
5, West; 6, Inner South; 7, Plains and Mountains; 8, Texas; 9, New England; 10, Midwest; 11, Florida; 12, California; 13, West Coast;
14, East Coast.
multidimensional scaling. Since these studies use preset areas, they have seldom been concerned with the
correlation between their findings and those of traditional dialectology, although that is not true of all
more contemporary work. Susan Tamasi (2003) developed a strategy in which respondents are given a
deck of cards representing the areas under consideration and are asked to stack them in as many piles as
necessary to show similarities.
A different approach to perception is realized in a
task in which respondents are asked to outline speech
areas on a blank map, label them with names of the
dialect and/or area and of typical speakers, and jot
down examples. Although respondent hand-drawn
maps are well known in cultural geography (e.g.,
Gould and White, 1974), there is no tradition of this
in dialectology. Several studies of the comments written on such maps have been carried out in the United
States and Japan (Preston, 1999; Long and Preston,
2002).
Preston and Howe (1987) developed a technique
for the computerized generalization of numbers of
such hand-drawn maps. Each is traced onto a digitizing pad that feeds the outline information into a
program. For each respondents identification of an
area, the program records one tally for each pixel
enclosed in or touched by the respondents boundary.
This technique allows the creation of composite maps
based on large numbers of respondents individual
ones. This procedure also allows questions other
than that of the best generalization to be asked: for
example, questions such as (1) where is the core of a
region?, (2) do different percentages of respondent
agreement show concentric patterns or do irregularities suggest alternative interpretations?, or (3) are
Figure 2 Mean scores of southeastern Michigan correctness ratings; higher scores indicate greater correctness.
dialectology, respondents have been asked more direct questions (Where are the correct and pleasant
varieties spoken?). Such ranking procedures have a
long history in cultural geography; Figure 2 is a map
of means scores for the correct task from southeastern Michigan and Figure 3 is a map from southern
(Alabama) respondents. These two maps show that
for both groups the regions most definitely associated
with incorrect English are the South and the New
York City area; they are the only sites that have
mean scores within the range 4.004.99. At the
other end of the scale, predictions about linguistic
security seem to be borne out. Michigan raters see
themselves as the only state in the 8.008.99 range,
exposing considerable linguistic self-confidence.
Alabama respondents, however, rate themselves in
the mediocre 5.005.99 range and clearly regard
other areas (Washington, D.C. and Maryland) as
superior. These correctness ratings show the predicted differences between the Michigan secure and
Alabama insecure raters and confirm the low prestige
assigned to southern and New York City varieties.
Figures 4 and 5 depict the ratings of the same
respondents for pleasant speech. The suggestion by
Giles and associates that local speech is affectively
preferred is strongly confirmed. Alabama respondents rate only the home state in the 8.008.99
range for pleasantness, and the Michigan raters
put only Washington, Colorado, and neighboring
Minnesota in the same 7.007.99 range along with
their home site.
Stereotypes and caricatures are definitively cataloged through such tasks and are only hinted at in
Table 1 The two factor groups from the ratings of all areasa
Factor group #1
descriptors
Weight
Factor group #2
descriptors
Weight
Smart
Educated
Normal
Good English
No drawl
No twang
Casual [Formal]
Fast
Down-to-earth
[Snobbish]
0.76
0.75
0.65
0.63
0.62
0.57
0.49
0.43
0.32
Polite
Friendly
Down-to-earth
(Normal)
(Casual)
0.74
0.74
0.62
(0.27)
(0.27)
a
From areas shown in Figure 1 (see text for discussion).
Descriptors in parentheses fall within the 0.250.29 rating
range; the negative loadings in Factor group #1 require these
two items to be interpreted at the opoosite polarity, shown in
square brackets to the left.
Factor
Mean
Attribute
Rank
Rank
Factor
Mean
Attribute
1 and 2
2
2 and 1
2
0a
1 and 2
1
1
1
1
1
1
4.66
4.58
4.54
4.20
4.09
3.22b
3.04b
2.96b
2.86b
2.72b
2.42b
2.22b
Casual
Friendly
Down-to-earth
Polite
Not nasal
Normal [Abnormal]
Smart [Dumb]
No twang [Twang]
Good English [Bad English]
Educated [Uneducated]
Fast [Slow]
No drawl [Drawl]
1
2
3
4
5
6
7
8
9
10
11
12
12
9.5
6
9.5
11
3
4
2
5
8
7
1
1 and 2
2
2 and 1
2
0
1 and 2
1
1
1
1
1
1
3.53
4.00
4.19
4.00
3.94
4.94
4.53
5.07
4.41
4.09
4.12
5.11
Casual
Friendly
Down-to-earth
Polite
Not nasal
Normal
Smart
No twang
Good English
Educated
Fast
No drawl
The two rows shown in boldface indicate the only significant (0.05) break between any two adjacent means scores.
Means values below 3.5 may be interpreted as the opposite polarity (attributes in square brackets).
the three southernmost voices, but strongly distinguished them from all others. The remaining middle
and northern voices were not accurately placed along
the northsouth dimension by either group of respondents, although their errors were very similar. These
findings continue to confirm the salience of southern
speech in the United States, whether from place-label
triggered caricature or from the presentation of actual
speech samples.
In its most recent phase, perceptual dialectology
has joined forces with sociophonetics in examining
the linguistic elements that influence dialect perception. Although eliciting folk imitations is one way
of approaching this problem (e.g., Preston, 1992;
and see the work of Betsy Evans, in Long and
Preston (2002)), the presentation of specific elements
(by name, by actual sample, or by computer-modified
samples) for identification, placement, and evaluation by respondents is perhaps a better way to grasp
even greater details of the triggering mechanisms of
language regard among the folk.
William Labov asked respondents in New York
City to evaluate the social status of speakers on the
basis of the frequency of nonprevocalic /r/ deletion
or stop substitution for the interdental fricatives,
providing an early example of perceptual (social)
dialectology. Another group led by Labov instrumentally manipulated the onset of the /aw/ diphthong of
an African-American speaker from Philadelphia (in
which the /a/ portion was fronted to a position nearer
//); they succeeded in showing that this fronting
alone was enough to signal white ethnicity to both
African- and European-American local respondents.
Cynthia Clopper played speech samples (with
appropriate regional phonetic features, e.g., New
England r-lessness) for Indiana University students.
attention of dialectologists, sociolinguists, and students of the social psychology of language. The work
done so far in this enterprise, however, shows that
perceptual dialectology is inseparable from affective
social factors in the speech community, and tells us
again that in folk linguistics in general the dominating
concerns appear to be pre- (and pro-)scription.
See also: Accent; Cuba: Language Situation; Dialects:
Early European Studies; English in the Present Day
(since ca. 1900); Folk Linguistics; France: Language Situation; Hungarian; Italian; Japan: History of Linguistics;
Language Attitudes; Language Education: Correctness
and Purism; Language Ideology; Mali: Language Situation; Netherlands: Language Situation; Prestige, Overt
and Covert; Reflexivity in Sociolinguistics; Relativism; Russian Federation: Language Situation; Social Psychology
and Language; Society and Language: Overview; Sociophonetics; Spain: Language Situation; Speech Community;
Speech Technologies: Language Variation; Switzerland:
Language Situation; Turkey: Language Situation; United
Kingdom: Language Situation; United States of America:
Language Situation; Wales: Language Situation.
Bibliography
Canobbio S & Ianna`ccaro G (2000). Contriubuto per una
bibliografia sulla dialettologia percettiva. Atlante linguistico ed etnografico del Piemonte Occidentale, #5. Turin:
Edizioni dellOrso and Universita` degli Studi di Torino,
Dipartamento de Scienze del Linguaggio and Consiglio
Nazionale delle Ricerchel. [A bibliography of works
devoted principally to perceptual dialectology.]
Clopper C G (2004). Linguistic experience and the perceptual classification of dialect variation. Doctoral diss.
(unpubl.), Indiana University.
Daan J (1970). Dialekten. In Von randstad tot landrand.
Bijdragen en Mededelingen der Dialecten Commissie van
de Koninklijke Nederlandse Akademie van Wetenschappen te Amsterdam, XXXVII. Amsterdam: N. V. Noord,
Hollandsche Uitgevers Maatschappij. [Transl. as Dialects, in Preston (ed.) (1999), 930.]
DAgostino M (ed.) (2002). Percezione della spazio, spazio
della percezione: La variazione linguistica fra nuovi e vecchi strumenti di analisi. Materiali e Ricerche 10; Atlante
Linguistico della Sicilia. Palermo: Centro di Studi
Filologici e Linguistici Siciliani, Dipartimento di Scienze
Filologiche e Linguistiche, Facolta` di Lettere e Filosofia.
Dailey-OCain J (1997). Geographic and sociopolitical
influences on language ideology and attitudes towards
language variation in postunification Germany. Doctoral
diss. (unpubl.), University of Michigan (Ann Arbor).
Diercks W (1988). Mental maps: linguistisch-geographische
Konzepte. Zeitschrift fu r Dialektologie und Linguistik
55, 280305. [Transl. in Long and Preston (eds.) (2002),
5170.]
Goebel H (1995). Geolinguistische Mental maps: Zum
Problem der subjectiven Dialektverwandschaft (anhand
Plichta B & Preston D R (2003). The /ay/s have it: stereotype, perception, and region. Paper presented at
New Ways of Analyzing Variation 32, Philadelphia, PA,
October 912.
Preston D R (1986). Five visions of America. Language in
Society 15, 221240.
Preston D R (1988a). Change in the perception of language
varieties. In Fisiak J (ed.) Historical dialectology: regional and social. Berlin, New York, Amsterdam: Mouton De
Gruyter. 475504.
Preston D R (1988b). Methods in the study of dialect
perception. In Thomas A (ed.) Methods in dialectology.
Clevedon, Avon, and Philadelphia: Multilingual Matters.
373395.
Preston D R (1988c). Sociolinguistic commonplaces in
variety perception. In Ferrara K et al. (eds.) Linguistic
change and contact: NWAV-XVI. Austin, TX: University
of Texas, Department of Linguistics. 279292.
Preston D R (1989a). Standard English spoken here:
the geographical loci of linguistic norms. In Ammon U
(ed.) Status and function of language and language
varieties. Berlin and New York: Walter de Gruyter.
324354.
Preston D R (1989b). Perceptual dialectology. Dordrecht:
Foris.
Preston D R (1992). Talking Black and talking White:
a study in variety imitation. In Hall J, Doane N &
Ringler D (eds.) Old English and new. New York:
Garland. 327355.
Preston D R (1993). Folk dialectology. In Preston D R
(ed.) American dialect research. Amsterdam: Benjamins.
333377.
Preston D R (1996). Where the worst English is spoken. In
Schneider E (ed.) Focus on the USA. Amsterdam and
Philadelphia: Benjamins. 297360.
Preston D R (1997). The South: the touchstone. In
Bernstein C, Nunnally T & Sabino R (eds.) Language
variety in the South. Tuscaloosa and London: University
of Alabama Press. 311351.
Preston D R (ed.) (1999). A handbook of perceptual dialectology. Amsterdam: Benjamins. [A bibliography of works
devoted principally to perceptual dialectology.]
Preston D R & Howe G M (1987). Computerized generalizations of mental dialect maps. In Denning K M et al.
(eds.) Variation in language: NWAV-XV. Stanford: Department of Linguistics, Stanford University. 361378.
Rakerd B & Plichta B (2003). More on perceptions of /A/
fronting. Paper presented at New Ways of Analyzing
Variation 32, Philadelphia, PA, October 912.
Rensink W G (1955). Dialectindeling naar opgaven van
medewerkers. Mededelingen der Centrale Commissie
voor Onderzoek van het Nederlandse Volkseigen 7,
2023. [Transl. as Informant classification of dialects,
in Preston (ed.) (1999), 37.]
Sibata T (1959). Ho gen kyo kai no ishiki [Subjective
consciousness of dialect boundaries]. Gengo Kenkyu
36, 130. [Transl. as Consciousness of dialect boundaries, in Preston (ed.) (1999), 3962.]
Tamasi S L (2003). Cognitive patterns of linguistic perceptions. Doctoral diss., University of Georgia (Athens).
Thus defined, perfectivity is also involved in the perfect tenses (see Perfects, Resultatives, and Experientials), as in:
(2) by 5 o clock, John had written a letter
Thus defined, perfectivity is also involved in the perfect tenses (see Perfects, Resultatives, and Experientials), as in:
(2) by 5 o clock, John had written a letter
there may be contexts in which a formally underspecified tense may turn out to be actually ambiguous
in terms of its temporal and/or aspectual reading).
Thus, the merging of the general aspectological terminology with the terminology used in the international literature with reference to the grammar of
Slavic languages seems perfectly justified. (Note that
the labels used in the individual grammatical traditions differ; e.g., Russian soverssennyi vs. nesoverssennyi). The situation, however, is not that
simple. In fact, although the translation provided
for (10), with the perfective verb napisat, is undisputable, the actual interpretation of (9), with imperfective pisat, may depend on the speakers
communicative intentions. Disregarding further complications, suffice it to say that in some contexts (9)
may be interpreted as referring to a terminated letterwriting action that did not lead to actual completion.
Thus, although the verb in (9) is imperfective, the
aspectual interpretation may be either imperfective or
perfective, depending on whether the event is viewed
as terminated or nonterminated. Because of this, it
is useful to distinguish between termination (i.e.,
attainment of the terminal point of the event, implying perfectivity) and completion (i.e., telic completion, hence a fortiori perfectivity). To understand this
point, consider (11), which presents a terminated
event (because its terminal point is obviously envisaged due to the delimiting adverbials), without
yielding a completed action (because the painting
of the whole wall is not entailed, again due to
the temporal limitation imposed by the time adverbials).
(11) John painted a wall for 2 hours/between 2 and 4
oclock/until 4 oclock
By contrast, consider (12), which is both terminated (perfective view) and completed (telic attainment), and (13), which is imperfective and, thus, by
implication noncompleted (cf. (3)).
(12) John painted a wall in 2 hours
(13) John was painting a wall
It is important to realize that although a given predicate may be telic in its basic characterization, the
actual fulfillment of its telic nature depends on context. In fact, although the two predicates used in
examples (2), (3), (11), (12), and (13) are all telic,
their telic nature is actually fulfilled only in perfective
contexts, namely (2) and (12). The important generalization here is that completion (i.e., fulfillment of
telicity) implies perfectivity, whereas the reverse is not
true, as suggested by (11) despite the presence of a
telic predicate and by (4), which is based on an atelic
predicate, for which the idea of completion cannot
possibly arise.
Coming back to the Russian examples in (9) and
(10), the best way to make sense of the contrast is to
state that the meaning originally conveyed by the
perfective versus imperfective opposition, as implemented in the verbal lexicon of Slavic languages,
corresponds to the opposition telic versus atelic,
although the further evolution of Slavic languages
has largely (as in Russian) or slightly (as in Bulgarian)
blurred the initial picture. Indeed, the originally
actional meaning (see Aspect and Aktionsart) has
been to a larger or smaller extent superseded by aspectual connotations. Needless to say, this interpretation presupposes that the notions aspect and
actionality be kept apart, as suggested by several
scholars (Comrie, 1976; Bertinetto, 1986; Smith,
1991).
The Progressive
As previously noted, one major implementation
of imperfectivity is by the use of the progressive
aspect. English has a specialized periphrasis to express this meaning, as shown in (3), (6), (8), and
(13). Other languages may have generically imperfective tenses, such as the French imperfect in (5), possibly supplemented by dedicated periphrases, as is
again partly the case in French with the e tre en train
de infinitive construction (which, however, undergoes restrictions) and is definitely so in Spanish,
where the estar gerund construction is generously
employed alongside the imperfect and the present.
One difference between dedicated progressive
devices and general imperfective tenses is that the
latter may receive other interpretations than progressivity, most notably habituality. As to progressive
devices, they often derive from locative expressions
or, less frequently, from movement expressions; in
some cases, reduplication processes provide the formal means to convey this meaning (Bybee et al.,
1994).
Progressivity may be regarded as the quintessence
of imperfectivity, and indeed it is often used to exemplify this notion. In progressive sentences, such as the
ones already presented, the event is viewed in a specific stage of its development, as observed in the
comment about (6). We may call such stage the focalization point. Note that the normal course of the
event may be interrupted immediately after this point
(compare (6) with Yesterday, Mary was going to the
beach. But on the way she met John, so she never got
there). We can capture this feature by stating that in
progressive sentences the continuation of the event
beyond the focalization point is left indeterminate,
irrespective of the knowledge of the actual state of
affairs held by the speaker. With atelic predicates,
this does not cause any problem of conceptualization
because the event (by its very nature) cannot imply
completion. With telic predicates, on the other hand,
formalization difficulties arise. These lie in the fact
that we should be able to express the idea that the
specific stage viewed belongs to an event that, if
carried out to its natural end, would involve completion even in cases in which we know for sure that
completion could not possibly obtain (as in while the
man was crossing the street, he was hit by a lorry).
The solution to this apparent paradox has been
sought, broadly speaking, in the domain of modal
logic, the idea being that the continuation of the
event (with its possibly hypothetical completion) lies
in some possible world, everything else being
equal (Dowty, 1979; Landman, 1992; Bonomi,
1997). Other scholars have looked for an alternative
solution in the direction of actional coercion, under
the assumption that the progressive alters the nature
of telic predicates (Parsons, 1990), although this line
of reasoning may present problems (Bertinetto,
1997).
One remarkable feature of the progressive is that
dedicated periphrases cannot normally be combined
with stative verbs. Consider the sentence *John is
possessing a car. This restriction is not absolute, however, as shown by John is resembling more and more
his father. In addition, individual languages may be
relatively tolerant, as is the case in English and
Spanish with copular verbs. In such cases, nonperiphrastic predicates depict permanent situations,
whereas periphrastic ones refer to contingent
Typological Remarks
It is worth observing that general-purpose imperfective tenses, to the extent that they are characterized as
such in the grammar of individual languages, tend
to express all imperfective values, disregarding the
additional presence of specialized devices. Thus, in
the Romance languages the imperfect (imperfective
past) may be used in both progressive and habitual
contexts, despite the availability of progressive and
habitual periphrases. This shows that the different
imperfective values share some important feature,
presumably to be sought in the idea of indeterminacy
(which in the case of progressivity may be identified
with the indeterminate continuation of the event,
whereas further assumptions need to be made for
habituality; cf. Lenci and Bertinetto, 2000).
In regard to markedness, the issue of the relative
hierarchy of perfectivity and imperfectivity has often
been raised. Both members of this pair have, however,
been proposed as the unmarked member by different
authors, with reference to different language families.
Indeed, if contextual flexibility and zero marking are
taken as the basic criteria for unmarkedness, then we
might observe that, for example, imperfectivity is the
unmarked member in, Chasu (Asu; Bantu), whereas
the reverse is to be observed in Obolo (Niger-Congo),
in which the imperfective tenses are obtained by adding the morpheme -ki- to perfective ones. It is not
easy, at the moment, to state which of these two
situations is more frequently observed in the languages of the world. Presumably, both situations
Bibliography
Bache C (1997). The study of aspect, tense and action:
towards a theory of the semantics of grammatical categories. Frankfurt am Main, Germany: Lang.
Bertinetto P M (1986). Tempo, aspetto e azione nel verbo
italiano. Il sistema dellindicativo. Firenze, Italy: Accademia della Crusca.
Bertinetto P M (1994). Statives, progressives and habituals:
analogies and divergences. Linguistics 32, 391423.
Bertinetto P M (1997). Il dominio tempo-aspettuale.
Demarcazioni, intersezioni, contrasti. Torino, Italy:
Rosenberg & Sellier.
Binnick R I (1991). Time and the verb: a guide to tense and
aspect. New York: Oxford University Press.
Bonomi A (1997). The progressive and the structure of
events. Journal of Semantics 14, 173205.
Boogaart R (1999). Aspect and temporal ordering. A
contrastive analysis of Dutch and English. The Hague:
Holland Academic Graphics.
Bybee J, Perkins R & Pagliuca W (1994). The evolution of
grammar. Tense, aspect, and modality in the languages of
the world. Chicago, IL: University of Chicago Press.
Cohen D (1989). Laspect verbal. Paris: Presses Universitaires Franc aises.
Comrie B (1976). Aspect. Cambridge, UK: Cambridge
University Press.
(ed.) (2000). Tense and aspect in the languages of
Dahl O
Europe. The Hague: Mouton-De Gruyter.
(2001). Languages without tense and aspect. In
Dahl O
Ebert K H & Zu n iga F (eds.) Arbeiten des Seminars
fu r allgemeine Sprachwissenschaft 16: Aktionsart and
aspectotemporality in non-European languages. Zu rich:
Universita t Zu rich. 159174.
Delfitto D (2002). Genericity in language: issues of syntax,
logical form and interpretation. Alessandria, Italy:
LOrso.
Dowty D (1979). Word meaning and Montague Grammar.
Dordrecht: Reidel.
Forsyth J (1970). A grammar of aspect: usage and meaning in the Russian verb. Cambridge, UK: Cambridge
University Press.
Garca Fernandez L & Camus Bergareche B (eds.) (2004).
El prete rito imperfecto. Madrid: Gredos.
Landman F (1992). The progressive. Natural Language
Semantics 1, 132.
Lenci A & Bertinetto P M (2000). Iterativity vs. habituality: on the iterative interpretation of perfective sentences.
In Higginbotham J, Pianesi F & Varzi A C (eds.) Speaking
of events. New York/Oxford: Oxford University Press.
245287.
all linguists consider it to be a cross-linguistic category at all; in their opinion, the perfects in different
languages are only linked by their common name,
reflecting the vicissitudes of the scholarly history of
traditional grammar. It was Dahl (1985: 129153)
who first showed that a cross-linguistic category
of perfect can be identified empirically, without a
preconceived definition of its semantics: The perfects
of various languages, different in their peripheral
uses, center around certain prototypical uses in a
nonrandom fashion. On the basis of Dahls results, a
questionnaire was developed that operationalizes
the definition of this gram (Lindstedt et al., 2000;
Lindstedt 2000): A language possesses a perfect if it
has a gram, associated with the verb, that is used
in the translation equivalents of most of the first
seven examples, illustrating different kinds of CR of
past situations, but is not used in the following four
examples in the questionnaire, consisting of short
narratives.
The perfect thus defined expresses present relevance and is also called the present perfect because
in various languages it has formal counterparts on
other temporal levels: these are the past perfects
(or pluperfects), future perfects (or futura exacta), and even past future perfects. As illustrations
we may take the English sentences she had read this
book, she will have read this book, and she
would have read this book, respectively (notice
that the last of these also carries modal meanings).
Although the term perfect may refer to all of these
grams, especially in studies considering the perfect to
be an aspect rather than a tense, it should be noted
that on nonpresent temporal levels, the notion of
continuing relevance is not so crucial: These other
perfects could simply be described as absolute-relative tenses that express temporal location of events
relative not only to the present time but also to each
other.
The distinction between resultatives and perfects
was established in linguistics in the 1980s, largely
owing to the important collective work edited by
Nedjalkov (1988). Resultatives signal that a state
exists as a result of a past action (Bybee et al., 1994:
54). The diagnostic difference between resultatives
all linguists consider it to be a cross-linguistic category at all; in their opinion, the perfects in different
languages are only linked by their common name,
reflecting the vicissitudes of the scholarly history of
traditional grammar. It was Dahl (1985: 129153)
who first showed that a cross-linguistic category
of perfect can be identified empirically, without a
preconceived definition of its semantics: The perfects
of various languages, different in their peripheral
uses, center around certain prototypical uses in a
nonrandom fashion. On the basis of Dahls results, a
questionnaire was developed that operationalizes
the definition of this gram (Lindstedt et al., 2000;
Lindstedt 2000): A language possesses a perfect if it
has a gram, associated with the verb, that is used
in the translation equivalents of most of the first
seven examples, illustrating different kinds of CR of
past situations, but is not used in the following four
examples in the questionnaire, consisting of short
narratives.
The perfect thus defined expresses present relevance and is also called the present perfect because
in various languages it has formal counterparts on
other temporal levels: these are the past perfects
(or pluperfects), future perfects (or futura exacta), and even past future perfects. As illustrations
we may take the English sentences she had read this
book, she will have read this book, and she
would have read this book, respectively (notice
that the last of these also carries modal meanings).
Although the term perfect may refer to all of these
grams, especially in studies considering the perfect to
be an aspect rather than a tense, it should be noted
that on nonpresent temporal levels, the notion of
continuing relevance is not so crucial: These other
perfects could simply be described as absolute-relative tenses that express temporal location of events
relative not only to the present time but also to each
other.
The distinction between resultatives and perfects
was established in linguistics in the 1980s, largely
owing to the important collective work edited by
Nedjalkov (1988). Resultatives signal that a state
exists as a result of a past action (Bybee et al., 1994:
54). The diagnostic difference between resultatives
and the perfect continuing, also known as the perfect of persistent situation, as in:
(6) Mary has been waiting for him for an hour.
Bibliography
(1989). The creation of tense and aspect
Bybee J & Dahl O
systems in the languages of the world. Studies in Language 13, 51103.
Bybee J, Perkins R & Pagliuca W (1994). The evolution of
grammar: tense, aspect, and modality in the languages of
the world. Chicago: University of Chicago Press.
Comrie B (1976). Aspect. Cambridge: Cambridge University Press.
Comrie B (1985). Tense. Cambridge: Cambridge University
Press.
(1985). Tense and aspect systems. Oxford: Basil
Dahl O
Blackwell.
(ed.) (2000). Tense and aspect in the languages of
Dahl O
Europe. Empirical Approaches to Language Typology,
EUROTYP 2026. Berlin: Mouton de Gruyter.
Speech addressed to others always involves a performance that offers the speaker an opportunity to perform well but simultaneously renders the speaker
vulnerable to demonstrating inadequacy, in particular
to losing face (Goffman, 1967; Brown and Levinson,
1987). Clearly, there are factors that affect the speakers
ability to speak in a manner that will be judged to yield
an adequate performance. What are these factors?
Modes of Discourse
Different demands are made on the speaker by different
modes of discourse (mode is borrowed from Smith,
2003). Modes involve types of discourse structuring,
each of which can be found in many different genres.
Any mode can be rendered cognitively simpler or more
difficult, and hence can impose fewer or more performance constraints on the speaker, by manipulating a
relatively small range of parameters (Brown, 1995:
4352), among them:
(a) the number and distinguishability of referents: it is
easier to instruct a listener to wire a 3-pin plug than
to wire a heating control panel, or to tell a story
involving a man, a woman, and a dog than a story
involving three tall, thin, anonymous men who
have no obvious distinguishing features.
(b) the relative complexity of spatial relationships: it
is easier to describe how to pose one brick directly
on another than how to cantilever the upper brick
over the lower one, or to describe an accident
involving a bus and a lorry on a straight road
Bibliography
Biber D, Johansson S, Leech G et al. (1999). Longman
grammar of spoken and written English. Harlow:
Longman.
Brown G (1995). Speakers, listeners and communication:
explorations in discourse analysis. Cambridge: Cambridge
University Press.
Brown P & Levinson S (1987). Politeness: some universals
in language usage. Cambridge: Cambridge University
Press.
Cicourel A (1981). Language and the structure of belief in
medical communication. Studia Linguistica 35(12),
7185.
Clark H H (1998). Communal lexicons. In Malmkjr K
& Williams J (eds.) Context in language learning
and language understanding. Cambridge: Cambridge
University Press. 6387.
Goffman E (1967). Interaction ritual. New York: Anchor
Books.
Smith C S (2003). Modes of discourse. Cambridge: Cambridge
University Press.
Performance in Culture
D Kapchan, New York University, New York, NY, USA
2006 Elsevier Ltd. All rights reserved.
Two Scenarios
An herbalist sits on a blanket that he has laid on the
asphalt in the main square, Jma al-Fna, in Marrakech,
Morocco (Figure 1). This square is known for its
performance traditions. In the evenings there are jugglers, storytellers, clairvoyants, acrobats, and snake
charmers. The audience for these performances is
largely Moroccan and the language used in the square
is Moroccan Arabic. During the day, however, there
are fewer performers. Abdelnacer and his brother
Abderrahman sit under a large umbrella in the square
waiting for customers (Figure 2). They are Saharaoui
(people from the Sahara), but they make their livelihood in Marrakech selling medicinal herbs, potions
against magic, and other ritual goods lizard skins,
amber and other resins used as incense, ostrich eggs,
whale bone, etc. (Figure 3). I sit with them in the
Bibliography
Biber D, Johansson S, Leech G et al. (1999). Longman
grammar of spoken and written English. Harlow:
Longman.
Brown G (1995). Speakers, listeners and communication:
explorations in discourse analysis. Cambridge: Cambridge
University Press.
Brown P & Levinson S (1987). Politeness: some universals
in language usage. Cambridge: Cambridge University
Press.
Cicourel A (1981). Language and the structure of belief in
medical communication. Studia Linguistica 35(12),
7185.
Clark H H (1998). Communal lexicons. In Malmkjr K
& Williams J (eds.) Context in language learning
and language understanding. Cambridge: Cambridge
University Press. 6387.
Goffman E (1967). Interaction ritual. New York: Anchor
Books.
Smith C S (2003). Modes of discourse. Cambridge: Cambridge
University Press.
Performance in Culture
D Kapchan, New York University, New York, NY, USA
2006 Elsevier Ltd. All rights reserved.
Two Scenarios
An herbalist sits on a blanket that he has laid on the
asphalt in the main square, Jma al-Fna, in Marrakech,
Morocco (Figure 1). This square is known for its
performance traditions. In the evenings there are jugglers, storytellers, clairvoyants, acrobats, and snake
charmers. The audience for these performances is
largely Moroccan and the language used in the square
is Moroccan Arabic. During the day, however, there
are fewer performers. Abdelnacer and his brother
Abderrahman sit under a large umbrella in the square
waiting for customers (Figure 2). They are Saharaoui
(people from the Sahara), but they make their livelihood in Marrakech selling medicinal herbs, potions
against magic, and other ritual goods lizard skins,
amber and other resins used as incense, ostrich eggs,
whale bone, etc. (Figure 3). I sit with them in the
Figure 2 Abdelnacer and Abderrahman with their fathers scale in Jma al-Fna, Marrakech.
radically over the centuries, the idea that ones environment shapes and influences ones actions has
remained consistent. (In 1871 Taylor defined culture
as that complex whole which includes knowledge,
belief, art, law, morals, customs, and any other capabilities and habits acquired by man as a member of
society. Boas introduced the notion of relativistic
culture the idea that all cultures are equally complex while Geertz defined culture as a model of
and for action.) Humanistic anthropologists such
as Abu-Lughod (1993) resist the idea of any kind of
cultural homogeneity that may prejudice the observer
to what is and is not there, encouraging instead a
writing against culture. One response to the difficulties of defining culture as a category is to always
analyze performance in its specificity, that is, in a
particular context one that is informed by historical
and political circumstance, by social and personal
environments whether of ethnicity, class, or gender.
Performances like the ones above are infused with
historical practices. Reading performances over time
is also reading history. This is clear in the analysis of
the two scenarios presented. In the first, Abdelnacer
performed his identity as an herbalist, as a Saharan,
Despite the fact that there exist models for behavior, actual performances are never the same twice.
There is always an emergent aspect, an indeterminate
element that changes each time (Bauman, 1977).
Authority is challenged (the herbalist in Paris sells
his goods despite prohibition) and generic convention
rebuked, reworked, and redefined.
Bauman noted the importance of the emergent in
verbal art in 1977, stating that completely novel and
completely fixed texts represent the poles of an ideal
continuum . . . between the poles lies the range of
emergent text structures to be found in empirical
performance (1977: 40). In his later work, fixed
texts became those that try to closely imitate a historical precedent in what he called, following Hymes, an
act of traditionalization (Bauman, 1992). Texts that
distinguish themselves from prior texts make a claim
to novelty or hybridity (Kapchan, 1996). The emergent in cultural phenomena, however, is what is unpredictable. Focusing on the emergent in situated
expression means identifying the points at which performances challenge or play with tradition in order
to inscribe a politics of difference onto the cultural
landscape.
Bibliography
Performative Clauses
K Allan, Monash University, Victoria, Australia
2006 Elsevier Ltd. All rights reserved.
to perform an action. There are five necessary conditions (NCs) and one sufficient condition (SC) on
using performatives felicitously.
To issue such an utterance is to perform the action
an action, perhaps, that we scarcely could perform, at
least with so much precision, in any other way. Here
are some examples (Austin, 1963: 22).
Performative Clauses
K Allan, Monash University, Victoria, Australia
2006 Elsevier Ltd. All rights reserved.
to perform an action. There are five necessary conditions (NCs) and one sufficient condition (SC) on
using performatives felicitously.
To issue such an utterance is to perform the action
an action, perhaps, that we scarcely could perform, at
least with so much precision, in any other way. Here
are some examples (Austin, 1963: 22).
Necessary Condition 2
In saying I promise in (3), the speaker makes a promise; but the words I promised in (4) do not constitute
the making of a promise; instead, they report that a
promise was made.
Necessary Condition 3
Uttered in felicitous circumstances, (7) has the illocutionary point of a bet, so the hearer can justifiably
reply Youre on! thereby taking up the bet and expecting the speaker to pay up when she or he loses, or vice
versa.
Necessary Condition 4
NC4 also places constraints on the modal auxiliaries that may occur in performative clauses.
(11) I will hereby promise to visit you next time
Im in town
Other Issues
Explicit performatives can be negative. Example (24)
performs an act of not-promising; note the scope of
the negative: An act of not-promising is different from
an act of promising not to do something, as in (25).
(24) I dont promise to come to your party, but Ill try
to make it.
Austin (1963, 1975) insisted on a distinction between what he called constatives, which have truth
values, and performatives, which, instead of truth values, have felicity conditions. In his opinion, (26) has
no truth value but is felicitous if there is a cat such
that the speaker has the ability and intention to put it
out, and it is infelicitous but not false otherwise.
This contrasts with (27), which is either true if the
speaker has put the cat out or false if he or she has
not.
(26) I promise to put the cat out.
(27) Ive put the cat out.
Interrogative
go!
I said go
*I said that go
Declarative
Declarative
Performative
Periphrasis 287
Speech Acts.
Bibliography
Allan K (1986). Linguistic meaning (vols. 12). London:
Routledge and Kegan Paul.
Austin J L (1963). Performative-constative. In Caton C E
(ed.) Philosophy and ordinary language. Urbana, IL:
University of Illinois Press. 2254. (Reprinted in Searle
J R (ed.) (1971). The philosophy of language. London:
Oxford University Press. 112.)
Austin J L (1975). How to do things with words (2nd edn.).
Urmson J O & Sbisa` M (eds.). Oxford: Oxford University
Press.
Bach K (1975). Performatives are statements too. Philosophical Studies 28, 229236. (Reprinted, slightly
amended, Bach K & Harnish R M (eds.) (1979). Linguistic communication and speech acts. Cambridge, MA:
MIT Press. 203208.)
Bach K & Harnish R M (1979). Linguistic communication
and speech acts. Cambridge, MA: MIT Press.
Ballmer T T & Brennenstuhl W (1981). Speech act classification: a study in the lexical analysis of English speech
activity verbs. Berlin: Springer-Verlag.
Cohen L J (1964). Do illocutionary forces exist? Philosophical Quarterly 14(55), 118137. (Reprinted in
Periphrasis
A Spencer, University of Essex, Colchester, UK
2006 Elsevier Ltd. All rights reserved.
Introduction
The term periphrasis is most commonly used to
denote a construction type in which a grammatical
property or feature is expressed by a combination of
words rather than a single (inflected) word form.
Periphrasis is generally applied to functional categories that are integrated into the inflectional system.
However, derivational processes for the creation of
new lexemes also show systematic regularities and,
where sufficiently regular, may even demonstrate a
paradigmatic organization (see below). A simple instance of periphrasis in English is the progressive and
perfect aspect construction expressed by the auxiliary
verbs BE, HAVE, and the -ing/-en forms, respectively, of
the lexical verb: The girls were singing/have sung.
Periphrasis 287
Speech Acts.
Bibliography
Allan K (1986). Linguistic meaning (vols. 12). London:
Routledge and Kegan Paul.
Austin J L (1963). Performative-constative. In Caton C E
(ed.) Philosophy and ordinary language. Urbana, IL:
University of Illinois Press. 2254. (Reprinted in Searle
J R (ed.) (1971). The philosophy of language. London:
Oxford University Press. 112.)
Austin J L (1975). How to do things with words (2nd edn.).
Urmson J O & Sbisa` M (eds.). Oxford: Oxford University
Press.
Bach K (1975). Performatives are statements too. Philosophical Studies 28, 229236. (Reprinted, slightly
amended, Bach K & Harnish R M (eds.) (1979). Linguistic communication and speech acts. Cambridge, MA:
MIT Press. 203208.)
Bach K & Harnish R M (1979). Linguistic communication
and speech acts. Cambridge, MA: MIT Press.
Ballmer T T & Brennenstuhl W (1981). Speech act classification: a study in the lexical analysis of English speech
activity verbs. Berlin: Springer-Verlag.
Cohen L J (1964). Do illocutionary forces exist? Philosophical Quarterly 14(55), 118137. (Reprinted in
Periphrasis
A Spencer, University of Essex, Colchester, UK
2006 Elsevier Ltd. All rights reserved.
Introduction
The term periphrasis is most commonly used to
denote a construction type in which a grammatical
property or feature is expressed by a combination of
words rather than a single (inflected) word form.
Periphrasis is generally applied to functional categories that are integrated into the inflectional system.
However, derivational processes for the creation of
new lexemes also show systematic regularities and,
where sufficiently regular, may even demonstrate a
paradigmatic organization (see below). A simple instance of periphrasis in English is the progressive and
perfect aspect construction expressed by the auxiliary
verbs BE, HAVE, and the -ing/-en forms, respectively, of
the lexical verb: The girls were singing/have sung.
288 Periphrasis
dojde Ivan
come Ivan
(1b) Ivan da dojde
Ivan DA come
Ivan should come!
DA
(2a) Az ste
pisa
pismoto
I
FUT
write.1SG the.letter
I will write the letter
(2b) te ste
pisat
pismoto
they FUT write.3PL the.letter
they will write the letter
(2c) az
ste
sam
napisal
pismoto
I
FUT
be.1SG write.SG the.letter
I will have written the letter
Periphrasis 289
(2d) te
s te
sa
napisali pismoto
they FUT be.3PL write.PL the.letter
they will have written the letter
In the simple future tenses, the s te particle is invariable (though historically it derives from a verb meaning want). However, there is a related function word
that takes person/number inflections proper to the
imperfect tense. This form combines with a da-clause
to give a kind of future-in-the-past construction,
generally interpreted as a conditional mood form:
(3a) az s tjax
da pis a
pismoto
I
S TA.1SG
DA
write.1SG the.letter
I would write the letter
(3b) te
s tjaxa
da pis at
pismoto
they S TA.3PL DA write.3PL the.letter
they would write the letter
(3c) az s tjax
da sa m
napisal
pismoto
I
S TA.1SG
DA
be.1SG write.SG the.letter
I would have written the letter
(3d) te
s tjaxa da sa
napisali pismoto
they S TA.3PL DA be.3PL write.PL the.letter
they would have written the letter
To negate the future-in-the-past or conditional construction, we use the 3sg negative imperfect form of
IMA, njamas e (glossed NJAMAS E), with the DA-clause.
Thus, corresponding to (3) we have:
(7a) az
njamas e da s a
pismoto
I
NJAMAS E
DA
write.1SG the.letter
I would not write the letter
(7b) te
njamas e da pis at
pismoto
they NJAMAS E DA write.3PL the.letter
they would not write the letter
(7c) az
njamas e da sa m
napisal pismoto
I
NJAMAS E
DA be.1SG write.SG the.letter
I would not have written the letter
(7d) te
njamas e da sa
napisali pismoto
they NJAMAS E DA be.3PL write.PL the.letter
they would not have written the letter
290 Periphrasis
Haspelmath is tacitly assuming (somewhat controversially) that perfective and imperfective verbs constitute two distinct (derivational?) classes rather than
two series of inflected forms of a single lexeme.
dadam uc ebnikot
I.give
the.textbook
I will not give him the textbook
NEG
ke
mu
go
FUT
TO.HIM
IT
Periphrasis 291
uc ebnika t
textbook.DEF
go
IT
dam
give.1SG
uc ebnika t
textbook.DEF
ovu knjigu
this book
c u
1SG.AUX
c itati
read.INF
(13b) c itati
c u
ovu knjigu
read.INF 1SG.AUX this book
I will read this book
The form c itac u now has to be analyzed as a synthetic word form. Finally, if we ask a yes-no question
in the future, then a simple answer will require just
the future auxiliary without a lexical verb. Because
the clitic/affix form would require a lexical verb to
attach to, such a one-word answer has to consist of
the full form of the auxiliary, as seen in (15):
(15a) c itac es
ovu knjigu?
read.FUT.2SG this book
will you read this book?
(15b) hoc u
1SG.AUX
yes (I will)
These compounds are productive and semantically transparent, and the lexical verb component
is accessible to morphosyntactic processes, such
as soo si-anaphora (equivalent to English do so)
and honorification with the o-VERB-ni naru construction, which are not possible with lexicalized compounds.
(17) verbal anaphora
soo si-hazimeru/tuzukeru
so do begin/continue
begin/continue to do so
292 Periphrasis
(18) honorifics
(18a) kaki-hazimeru
write-begin
(18b) o-kaki-ni nari hazimeru
HON-write-HON BECOME begin
begin to write
Derivational Periphrases
Although it is not the standard way of using the term,
periphrasis can also be applied to multiword lexical
structures such as particle verbs (make a story up),
light verb constructions (render assistance), and constructions in which the lexical stock is expanded
systematically by multiword combinations. Other
types include serial verb constructions (see Durie,
1997) and aspectual and causative complex predicates in languages such as Indo-Aryan (Butt, 1997)
and Romance (Alsina, 1997).
A well-known instance of lexical multiword combinations is the case of Germanic particle verb
constructions or phrasal verb (see the papers in
Dehe et al., 2002, for recent discussion). In some
instances, the combination has the appearance of a
normal, semantically compositional piece of syntax,
as in He pulled the cork out. Here, the word out
appears to be an intransitive preposition having the
function of an adverbial of location, much like the
Periphrasis 293
Conclusions
Periphrastic constructions differ from general syntactic constructions in that they realize a grammatical
property or derivational pattern. Various diagnostics
are being developed for distinguishing periphrastic
constructions from the rest of syntax. While remaining essentially syntactic in form, periphrases thus exhibit certain of the properties of the morphology and
may interact with morphological systems to the extent of being counted as part of the morphological
paradigm of a lexeme (Sadler and Spencer, 2001) or
of alternating with purely morphological constructions. Such patterning can be best understood by
appealing to a lexeme and its paradigm of forms,
294 Periphrasis
Bibliography
Ackerman F, Blevins J P & Stump G T (eds.) (Forthcoming). Periphrasis and paradigms. Stanford: CSLI
Publications.
Ackerman F & LeSourd P (1997). Toward a lexical representation of phrasal predicates. In Alsina et al. (eds.).
67106.
Ackerman F & Webelhuth G (1998). A theory of predicates.
Stanford University: Center for the Study of Language
and Information.
Ackerman F & Stump G T (2004). Paradigms and periphrastic expression: a study in realization-based lexicalism. In Sadler L & Spencer A (eds.) Projecting
morphology. Stanford University: Center for the Study
of Language and Information. 111158.
Alsina A (1997). Causatives in Bantu and Romance. In
Alsina et al. (eds.). 203246.
Alsina A, Bresnan J & Sells P (eds.) (1997). Complex
predicates. Stanford University: Center for the Study of
Language and Information.
Blevins J P (2001). Realisation-based lexicalism (F. Ackerman & G. Webelhuth, A Theory of Predicates). Journal
of Linguistics 37, 355365.
Blevins J P (Forthcoming). English inflection and derivation. In Aarts B & McMahon A M S (eds.) Handbook of
English linguistics. Oxford: Blackwell Publishers.
Booij G (2002). Separable complex verbs in Dutch: a case
of periphrastic word formation. In Dehe et al. (eds.).
2142.
Booij G (2003). Periphrastic word formation. In Booij G,
DeCesaris J, Ralli A & Scalise S (eds.) Topics in morphology. Selected Papers from the Third Mediterranean